Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
What a horrible comment. If both your parents due of cancer just before your first child is born (and I hope they don't/didn't) then you will realise how awful it is even for older people.
You seem to be a prat. Dying is awful, who knew? But fixing cancer has nothing like the moral and emotional impact of the great fix in puerperal mortality of the 19th and 20th Cs and why would it?
Proper sport tomorrow - not the nonsense at Nottingham but the class at Ascot, my favourite five days of the racing year. I'm a little disappointed they've tacked on an unnecessary final race but the great races are in place.
Picking the bones out of yesterday's first round in the French legislative elections - Ensemble finished fractionally ahead of NUPES but the latter had four candidates directly elected while Ensemble only had one.
Ensemble lead in 205 seats, NUPES are ahead in 194, National Rally in 110 and the UDC (the alliance of LR and other centre-right and centre parties) is head in 50 with Others ahead in 10.
Projections still suggest Ensemble will be close to or just over an overall majority (289) with NUPES on about 200 but we'll know a lot more this time next week.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
Moderna have an mRNA HIV vaccine in trials at the moment.
Can humanity eradicate a deadly disease, only 40 years after it was discovered?
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Like, Shakespeare's parents and teachers input some language and stories into him, and he processed it into a lorra plays and sonnets. Neat trick but there's no way any of these Stratford kids are sentient, because I arbitrarily and thoughtlessly stipulate they are not
It's not your impressionability which distinguishes you from the average punter. Trust me on this.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
Moderna have an mRNA HIV vaccine in trials at the moment.
Can humanity eradicate a deadly disease, only 40 years after it was discovered?
Could an achievement be more inherently unexciting, if it does?
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
Yes, I think people looking into it from the outside are quite impressed by what is essentially a bit of deep learning and "creative" thinking by an AI but none of these even come close to consciousness or sentience. I'm open to the idea that AI will reach that sooner rather than later but I also don't think people quote understand the implications of what an AI will do or how it will act.
But we don’t know what human consciousness or sentience really is, though some experimental evidence indicates the inner voice is merely an echo of an underlying function, rather than the driver of the train itself.
We soon run into a solipsist type argument, such that if a computer programme exhibits behaviours indistinguishable from a person with sentience, we will accept it has sentience. What’s the alternative?
And indeed that shift will probably occur sooner than that, given sentience is a) likely on a continuous scale rather than being binary, b) artificial sentience would differ dramatically from human sentience.
Throw enough neural connections together in the right way and you probably get there. What’s the threshold for that? We don’t know and we probably won’t realise it for a while when it happens. It’s honestly getting close to be a question for philosophers rather than computer scientists.
Proper sport tomorrow - not the nonsense at Nottingham but the class at Ascot, my favourite five days of the racing year. I'm a little disappointed they've tacked on an unnecessary final race but the great races are in place.
Picking the bones out of yesterday's first round in the French legislative elections - Ensemble finished fractionally ahead of NUPES but the latter had four candidates directly elected while Ensemble only had one.
Ensemble lead in 205 seats, NUPES are ahead in 194, National Rally in 110 and the UDC (the alliance of LR and other centre-right and centre parties) is head in 50 with Others ahead in 10.
Projections still suggest Ensemble will be close to or just over an overall majority (289) with NUPES on about 200 but we'll know a lot more this time next week.
Ensemble would expect to take almost all of around 200 seats where they led in the first round. The main battlegrounds are 133 seats where Ensemble is behind Nupes, and 65 where Ensemble is behind RN. Now of course many are unwinnable, but the point of the second round is to allow voters to rally behind the president (or not!).
"Appeal court judges have refused to stop the government's first flight taking asylum seekers to Rwanda on Tuesday."
Is it possible that the government never intended this to go ahead - implementation will be patchy, embarrassing and an epic fail at best - and assumed the courts would stop it so that they could blame the treacherous 'enemies of the people' judiciary?
If it stops the boats coming across the Channel then Priti will be lauded by the Tory right and the chances of her succeeding Boris as Tory leader will increase accordingly
Very polite and very serious question. Do you personally think this will stop the boats coming across the Chanel?
The opening up of regular ways of getting across the Channel will stop the boats coming.
No, it won’t
Yeah it will.
Getting into the UK used to be easy.
Then Covid hit.
Suddenly the only way in was a small boat.
Now Covid has receded, it's going to be massively easier to cross the channel more cheaply and safer. You know: you pay £250 for a spot in the back of someone's van.
Why the hell go by boat, when you're much less likely to get intercepted by the authorities?
Because the UK can and will stop you if you come by car or lorry. The boatniks have realised that if you arrive by boat there is nothing the UK Border people can do. We can’t push them back because they might drown, no UK government can risk this, no coastguard wants that on their conscience. We are powerless. So we just let them come, indeed we escort them to safely in the UK. And once they are here nearly all of them (all of them?) get to stay, because they have already dumped all ID and they therefore become “refugees’
The boats are not going to stop. I bet they will increase in number. The only solution is France toughening up (they won’t, they want them gone) the EU suddenly stopping all migrants at the EU border (they can’t) or something like Rwanda. The Australian solution
Are you high?
The UK inspects well under 1% of trucks and vans coming into the UK.
And you know what happens if you get caught in the back of the van? YOU CLAIM ASYLUM.
Remember: the goal isn't to get caught. It's to not get caught, and to enter the illegal economy with no record of your presence.
So why, oh Oracle, would one bother to clamber into a boat?
Because getting into a boat is a doddle compared to “finding a lorry”. No enormous fences to climb at thje Eurostar terminals, no lorry drivers determined to hunt you down (because they get heavily fined), no chance of being asphyxiated like those poor Vietnamese guys
And we keep saying how dangerous it is to cross the Channel but the evidence is mounting that it simply is not. How many are crossing every year? 40,000 now? And how many have drowned? A handful? The Channel is tricky because it is busy but that same busyness means there are lots of people who can pick you up
You have a less than 0.1% chance of drowning, maybe much less
The boats will keep coming. It is the easiest way to get into the UK
The boats didn't exist before Covid.
And the boats don't even really exist now for six-seven months of the year when the Channel is rough. With Covid receding, do you really think that people will hang around for half a year.
I'll do a bet with you if you like.
I reckon that there will be a 50% or greater drop in the number of people arriving by boat in the next 12 months, but that overall numbers of people claiming asylum in the UK will actually rise.
£100.
I’ll take a version of that bet but
1. Let’s keep it simple and
2. It feels properly wrong to bet and gain on suffering
So, let’s just make it “50% or greater drop in the number of people arriving by boat in the next 12 months”. (We need to set a commencement date)
And let’s make it £50 goes to a refugee charity of the winner’s choice? The loser can send the evidence of payment
Deal? (But we still need to frame the date)
Btw, I hope you’re right. I’d love to lose this as it means we have “solved” the boat problem, even if it is not truly solved. I just can’t see it happening
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
Proper sport tomorrow - not the nonsense at Nottingham but the class at Ascot, my favourite five days of the racing year. I'm a little disappointed they've tacked on an unnecessary final race but the great races are in place.
Picking the bones out of yesterday's first round in the French legislative elections - Ensemble finished fractionally ahead of NUPES but the latter had four candidates directly elected while Ensemble only had one.
Ensemble lead in 205 seats, NUPES are ahead in 194, National Rally in 110 and the UDC (the alliance of LR and other centre-right and centre parties) is head in 50 with Others ahead in 10.
Projections still suggest Ensemble will be close to or just over an overall majority (289) with NUPES on about 200 but we'll know a lot more this time next week.
Are you attending any of the days Stodge? Don’t forget a hat! 🌞
Ensemble would expect to take almost all of around 200 seats where they led in the first round. The main battlegrounds are 133 seats where Ensemble is behind Nupes, and 65 where Ensemble is behind RN. Now of course many are unwinnable, but the point of the second round is to allow voters to rally behind the president (or not!).
The question may be in how many of the 110 seats would the anti-RN vote coalesce enough around the second place candidate to win the seat?
I don't know in how many of the 110 seats NUPES are second or Ensemble second.
Beyond that, in how many of the seats where NUPES is leading would enough anti-left votes rally to say an Ensemble candidate to take the seat?
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
Moderna have an mRNA HIV vaccine in trials at the moment.
Can humanity eradicate a deadly disease, only 40 years after it was discovered?
Could an achievement be more inherently unexciting, if it does?
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace. 🗽
Nb. Having said that, I havn’t even got out my pyjama’s today 🤦♀️ And my New Year resolution to do charity work has not yet happened.
Former Manchester United and Chelsea chief executive Peter Kenyon is fronting a consortium who have entered talks to try to buy Everton.
Kenyon is part of a group that includes chief executive of Minneapolis-based Talon Real Estate Maciek Kaminski and American businessman John Thornton, and is being advised by investment specialist Michael Klein, along with the US law firm Weil, Gotshal and Manges.
It is understood heads of terms have been signed although talks are described as being at a “relatively early stage”, with owner Farhad Moshiri believed to value Everton in excess of £500million, taking into account the club’s debt.
The consortium may want guarantees that Everton will not face a points deduction or heavy fine over Financial Fair Play issues before entering into a legally-binding agreement, but the club have always insisted they have not broken regulations.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
The 5th district of Essonne is a good example of the battles. The Fields medallist Cédric Villani ratted from the presidential majority when he wasn't nominated for Paris mayor and recently went to Nupes. He won 15k votes, well ahead of Ensemble on 12k. but there are 6k Republican votes and 4k for the extreme right, so it is easy to imagine its going either way depending on voter mobilisation.
Are you attending any of the days Stodge? Don’t forget a hat! 🌞
When will you share with us a few tips?
Unfortunately, I have to work for a living so the tv is my only hope.
Selections for tomorrow's main races:
2.30: BAAEED 3.05: PERSIAN FORCE 3.45: MAN OF PROMISE (e/w) - my Aussie contracts tell me NATURE STRIP will win four lengths hard-held. 4.20: LUSAIL (e/w)
Proper sport tomorrow - not the nonsense at Nottingham but the class at Ascot, my favourite five days of the racing year. I'm a little disappointed they've tacked on an unnecessary final race but the great races are in place.
Picking the bones out of yesterday's first round in the French legislative elections - Ensemble finished fractionally ahead of NUPES but the latter had four candidates directly elected while Ensemble only had one.
Ensemble lead in 205 seats, NUPES are ahead in 194, National Rally in 110 and the UDC (the alliance of LR and other centre-right and centre parties) is head in 50 with Others ahead in 10.
Projections still suggest Ensemble will be close to or just over an overall majority (289) with NUPES on about 200 but we'll know a lot more this time next week.
I expect most LR-UDC voters to vote for Ensemble over NUPES or National Rally so Macron's party should get close to a majority or at it yes in the runoff
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
Moderna have an mRNA HIV vaccine in trials at the moment.
Can humanity eradicate a deadly disease, only 40 years after it was discovered?
Could an achievement be more inherently unexciting, if it does?
Well, I would be excited!
You, me, and a billion Africans, where AIDS is still endemic and the Western treatments unobtainable.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
Yes, I think people looking into it from the outside are quite impressed by what is essentially a bit of deep learning and "creative" thinking by an AI but none of these even come close to consciousness or sentience. I'm open to the idea that AI will reach that sooner rather than later but I also don't think people quote understand the implications of what an AI will do or how it will act.
But we don’t know what human consciousness or sentience really is, though some experimental evidence indicates the inner voice is merely an echo of an underlying function, rather than the driver of the train itself.
We soon run into a solipsist type argument, such that if a computer programme exhibits behaviours indistinguishable from a person with sentience, we will accept it has sentience. What’s the alternative?
And indeed that shift will probably occur sooner than that, given sentience is a) likely on a continuous scale rather than being binary, b) artificial sentience would differ dramatically from human sentience.
Throw enough neural connections together in the right way and you probably get there. What’s the threshold for that? We don’t know and we probably won’t realise it for a while when it happens. It’s honestly getting close to be a question for philosophers rather than computer scientists.
Yes. Quite. Some on PB seem unable to grasp this
Turing foresaw this problem of course. Hence the simple genius of the Turing test. If they can pass as intelligent, they are intelligent, for all intents and purposes
If that Google employee was who was suspended for believing his computer is “intelligent” and deserves “empathy” is not a hoax, then that computer has passed the Turing test
(And yes I know simple chat bots have passed it before blah blah)
Picking the bones out of yesterday's first round in the French legislative elections - Ensemble finished fractionally ahead of NUPES but the latter had four candidates directly elected while Ensemble only had one.
Ensemble lead in 205 seats, NUPES are ahead in 194, National Rally in 110 and the UDC (the alliance of LR and other centre-right and centre parties) is head in 50 with Others ahead in 10.
Projections still suggest Ensemble will be close to or just over an overall majority (289) with NUPES on about 200 but we'll know a lot more this time next week.
I expect most LR-UDC voters to vote for Ensemble over NUPES or National Rally so Macron's party should get close to a majority or at it yes in the runoff
Yes, that seems more than plausible. 300 for Macron, 200 for Melenchon looks about right - RN will lose a lot of the seats where they lead while UDC should hold most of the ones where they lead so maybe 15-20 for RN and 50 for the UDC parties.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
Moderna have an mRNA HIV vaccine in trials at the moment.
Can humanity eradicate a deadly disease, only 40 years after it was discovered?
Could an achievement be more inherently unexciting, if it does?
Well, I would be excited!
Quite so. Plus any other medic or nurse or paramedic having to deal with bodily fluids. For that matter, policemen too for instance. But also it's such good news if it passes trials, for everyone else, as Sandpit says.
PBers will be pleased to learn that long-standing territorial dispute over Hans Island aka Tartupaluk is over!
Concluded by agreement to split the small Arctic island in two between of Territory of Nunavut, and Greenland aka Kalaallit Nunaat, thus creating new international land frontier between Canada and Kingdom of Denmark.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
Yes, I think people looking into it from the outside are quite impressed by what is essentially a bit of deep learning and "creative" thinking by an AI but none of these even come close to consciousness or sentience. I'm open to the idea that AI will reach that sooner rather than later but I also don't think people quote understand the implications of what an AI will do or how it will act.
But we don’t know what human consciousness or sentience really is, though some experimental evidence indicates the inner voice is merely an echo of an underlying function, rather than the driver of the train itself.
We soon run into a solipsist type argument, such that if a computer programme exhibits behaviours indistinguishable from a person with sentience, we will accept it has sentience. What’s the alternative?
And indeed that shift will probably occur sooner than that, given sentience is a) likely on a continuous scale rather than being binary, b) artificial sentience would differ dramatically from human sentience.
Throw enough neural connections together in the right way and you probably get there. What’s the threshold for that? We don’t know and we probably won’t realise it for a while when it happens. It’s honestly getting close to be a question for philosophers rather than computer scientists.
Yes. Quite. Some on PB seem unable to grasp this
Turing foresaw this problem of course. Hence the simple genius of the Turing test. If they can pass as intelligent, they are intelligent, for all intents and purposes
If that Google employee was who was suspended for believing his computer is “intelligent” and deserves “empathy” is not a hoax, then that computer has passed the Turing test
(And yes I know simple chat bots have passed it before blah blah)
Except that isn't what he said or thought.....it was can machines do well in an imitation game. Nothing about being "intelligent".
In Alan Turing's 1950 paper, Computing Machinery and Intelligence, the mathematician posed the question: "Can machines think?" But almost immediately he dismissed that question as too "meaningless" to be worthy of discussion, and swapped it for the much-more specific: "Are there imaginable digital computers which would do well in the imitation game?"
Turing's original "imitation game" had nothing to do with artificial intelligence.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
Yes, I think people looking into it from the outside are quite impressed by what is essentially a bit of deep learning and "creative" thinking by an AI but none of these even come close to consciousness or sentience. I'm open to the idea that AI will reach that sooner rather than later but I also don't think people quote understand the implications of what an AI will do or how it will act.
But we don’t know what human consciousness or sentience really is, though some experimental evidence indicates the inner voice is merely an echo of an underlying function, rather than the driver of the train itself.
We soon run into a solipsist type argument, such that if a computer programme exhibits behaviours indistinguishable from a person with sentience, we will accept it has sentience. What’s the alternative?
And indeed that shift will probably occur sooner than that, given sentience is a) likely on a continuous scale rather than being binary, b) artificial sentience would differ dramatically from human sentience.
Throw enough neural connections together in the right way and you probably get there. What’s the threshold for that? We don’t know and we probably won’t realise it for a while when it happens. It’s honestly getting close to be a question for philosophers rather than computer scientists.
Yes. Quite. Some on PB seem unable to grasp this
Turing foresaw this problem of course. Hence the simple genius of the Turing test. If they can pass as intelligent, they are intelligent, for all intents and purposes
If that Google employee was who was suspended for believing his computer is “intelligent” and deserves “empathy” is not a hoax, then that computer has passed the Turing test
(And yes I know simple chat bots have passed it before blah blah)
The point being that the Turing Test is a necessary but not sufficient condition for 'intelligence" (something that Turing was not even claiming for things that won the Imitation Game) - not least because it is perfectly possible for humans to consistently fail the test, being misidentified by other humans as machines with statistical significance. And fooling one Google researcher with a vested interest is hardly significant.
We are closer than ever to general purpose artificial intelligence, and yet, also, miles away.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
Yes, I think people looking into it from the outside are quite impressed by what is essentially a bit of deep learning and "creative" thinking by an AI but none of these even come close to consciousness or sentience. I'm open to the idea that AI will reach that sooner rather than later but I also don't think people quote understand the implications of what an AI will do or how it will act.
But we don’t know what human consciousness or sentience really is, though some experimental evidence indicates the inner voice is merely an echo of an underlying function, rather than the driver of the train itself.
We soon run into a solipsist type argument, such that if a computer programme exhibits behaviours indistinguishable from a person with sentience, we will accept it has sentience. What’s the alternative?
And indeed that shift will probably occur sooner than that, given sentience is a) likely on a continuous scale rather than being binary, b) artificial sentience would differ dramatically from human sentience.
Throw enough neural connections together in the right way and you probably get there. What’s the threshold for that? We don’t know and we probably won’t realise it for a while when it happens. It’s honestly getting close to be a question for philosophers rather than computer scientists.
Yes. Quite. Some on PB seem unable to grasp this
Turing foresaw this problem of course. Hence the simple genius of the Turing test. If they can pass as intelligent, they are intelligent, for all intents and purposes
If that Google employee was who was suspended for believing his computer is “intelligent” and deserves “empathy” is not a hoax, then that computer has passed the Turing test
(And yes I know simple chat bots have passed it before blah blah)
See Arthur C. Clarke - any sufficiently advanced technology will appear to be magic. The victorians had chess playing automatons. Clearly a trick, but probably fooled some. With the Turing test, once the parameters are known, you can find a way to pass it.
Sentience is incredibly hard to prove, or even define. But simulating something that looks like is getting easier.
Today's comedy in the French election is the row over who won the popular vote. The media said Nupes until the official interior ministry figures said Ensemble, at which point all hell broke loose. It turned out a bunch of candidates' allegiances were debatable. Nupes in particular had been including a bunch of dissident Socialist Party members who wanted nothing to do with the extreme-left alliance. Predictably Mélenchon blamed the difference on the white devil in government. Cue Le Monde and others explaining that their lists were based on expert judgement, etc etc.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
But Yerevan does not feel like that. Trust me, I’m here. This feels like demotic prosperity. It’s not a few cats in German cars. It’s the whole city
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
Yes, I think people looking into it from the outside are quite impressed by what is essentially a bit of deep learning and "creative" thinking by an AI but none of these even come close to consciousness or sentience. I'm open to the idea that AI will reach that sooner rather than later but I also don't think people quote understand the implications of what an AI will do or how it will act.
But we don’t know what human consciousness or sentience really is, though some experimental evidence indicates the inner voice is merely an echo of an underlying function, rather than the driver of the train itself.
We soon run into a solipsist type argument, such that if a computer programme exhibits behaviours indistinguishable from a person with sentience, we will accept it has sentience. What’s the alternative?
And indeed that shift will probably occur sooner than that, given sentience is a) likely on a continuous scale rather than being binary, b) artificial sentience would differ dramatically from human sentience.
Throw enough neural connections together in the right way and you probably get there. What’s the threshold for that? We don’t know and we probably won’t realise it for a while when it happens. It’s honestly getting close to be a question for philosophers rather than computer scientists.
Yes. Quite. Some on PB seem unable to grasp this
Turing foresaw this problem of course. Hence the simple genius of the Turing test. If they can pass as intelligent, they are intelligent, for all intents and purposes
If that Google employee was who was suspended for believing his computer is “intelligent” and deserves “empathy” is not a hoax, then that computer has passed the Turing test
(And yes I know simple chat bots have passed it before blah blah)
The point being that the Turing Test is a necessary but not sufficient condition for 'intelligence" (something that Turing was not even claiming for things that won the Imitation Game) - not least because it is perfectly possible for humans to consistently fail the test, being misidentified by other humans as machines with statistical significance. And fooling one Google researcher with a vested interest is hardly significant.
We are closer than ever to general purpose artificial intelligence, and yet, also, miles away.
The fact Google themselves are saying its nonsense, sort of says even the vested interests aren't trying to claim that. Its one bloke who says the computer "spoke to me", in the way even highly qualified people have reported seeing ghosts, being abducted by aliens etc.
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
Former Manchester United and Chelsea chief executive Peter Kenyon is fronting a consortium who have entered talks to try to buy Everton.
Kenyon is part of a group that includes chief executive of Minneapolis-based Talon Real Estate Maciek Kaminski and American businessman John Thornton, and is being advised by investment specialist Michael Klein, along with the US law firm Weil, Gotshal and Manges.
It is understood heads of terms have been signed although talks are described as being at a “relatively early stage”, with owner Farhad Moshiri believed to value Everton in excess of £500million, taking into account the club’s debt.
The consortium may want guarantees that Everton will not face a points deduction or heavy fine over Financial Fair Play issues before entering into a legally-binding agreement, but the club have always insisted they have not broken regulations.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
We are now very good at building special-purpose problem solving machines for increasingly challenging domains (I'd say the AI that was capable of taking text-and-math descriptions of propositions and rewriting them into formal language for our automated solvers is a more interesting recent example).
But as I've said before, we are miles from general purpose intelligence.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Do you know how ball and player tracking works?
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
But Yerevan does not feel like that. Trust me, I’m here. This feels like demotic prosperity. It’s not a few cats in German cars. It’s the whole city
I’ve now walked for 2km and on it goes
So did Panama City 20 years ago. Massive modern high rises, fancy bars and restaurants, for several square kms.....while 30 mins away people lived in wooden shacks without shoes.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
This is true. Whenever I go to park my car's numberplate gets misread and I have to hit the button to talk to a human.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
But Yerevan does not feel like that. Trust me, I’m here. This feels like demotic prosperity. It’s not a few cats in German cars. It’s the whole city
I’ve now walked for 2km and on it goes
So did Panama City 20 years ago. Massive modern high rises, fancy bars and restaurants, for several square kms.....while 30 mins away people lived in wooden shacks without shoes.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
Yes, I think people looking into it from the outside are quite impressed by what is essentially a bit of deep learning and "creative" thinking by an AI but none of these even come close to consciousness or sentience. I'm open to the idea that AI will reach that sooner rather than later but I also don't think people quote understand the implications of what an AI will do or how it will act.
But we don’t know what human consciousness or sentience really is, though some experimental evidence indicates the inner voice is merely an echo of an underlying function, rather than the driver of the train itself.
We soon run into a solipsist type argument, such that if a computer programme exhibits behaviours indistinguishable from a person with sentience, we will accept it has sentience. What’s the alternative?
And indeed that shift will probably occur sooner than that, given sentience is a) likely on a continuous scale rather than being binary, b) artificial sentience would differ dramatically from human sentience.
Throw enough neural connections together in the right way and you probably get there. What’s the threshold for that? We don’t know and we probably won’t realise it for a while when it happens. It’s honestly getting close to be a question for philosophers rather than computer scientists.
Yes. Quite. Some on PB seem unable to grasp this
Turing foresaw this problem of course. Hence the simple genius of the Turing test. If they can pass as intelligent, they are intelligent, for all intents and purposes
If that Google employee was who was suspended for believing his computer is “intelligent” and deserves “empathy” is not a hoax, then that computer has passed the Turing test
(And yes I know simple chat bots have passed it before blah blah)
The point being that the Turing Test is a necessary but not sufficient condition for 'intelligence" (something that Turing was not even claiming for things that won the Imitation Game) - not least because it is perfectly possible for humans to consistently fail the test, being misidentified by other humans as machines with statistical significance. And fooling one Google researcher with a vested interest is hardly significant.
We are closer than ever to general purpose artificial intelligence, and yet, also, miles away.
The fact Google themselves are saying its nonsense, sort of says even the vested interests aren't trying to claim that. Its one bloke who says the computer "spoke to me", in the way even highly qualified people have reported seeing ghosts, being abducted by aliens etc.
You might say that Google the corporate has a vested interest in not supporting something that could be reputationally damaging if it is demonstrated to be nonsense. But yes, I agree.
Partnering with other companies and organisations, including... ESA. Although this makes sense for a load of reasons. (*) Actually they wanted to keep us out of various critical parts of it, so we left the whole program.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Do you know how ball and player tracking works?
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
Interestingly, it still isn't a fully solved problem. The likes of Statsbomb employ 100s of people in Egypt to manually click screens to fix up all the data. Also why offside still isn't automatic (although FIFA trialling something is that more automated, the people behind it still aren't promising fully auto).
And for sports with lots of occlusion and particularly players in similar uniforms / covered heads e.g. America Football, a really open problem.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
But Yerevan does not feel like that. Trust me, I’m here. This feels like demotic prosperity. It’s not a few cats in German cars. It’s the whole city
I’ve now walked for 2km and on it goes
So did Panama City 20 years ago. Massive modern high rises, fancy bars and restaurants, for several square kms.....while 30 mins away people lived in wooden shacks without shoes.
Well I’m intending to report back from the boondocks of Armenia soon. So I shall give my honest account! I expect it to be a lot poorer
Thing is with Panama you could see where the money was coming from, tho, no?
Drugs, tourism, shipping, the Canal, brass plate companies, money laundering
Where is the money coming from in Yerevan? Buggered if I know. But it is here. And good pick to them. They deserve it. What with all that genocide and all
I’ve decided the city it most reminds me of is Tel Aviv
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Do you know how ball and player tracking works?
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
Interestingly, it still isn't a fully solved problem. The likes of Statsbomb employ 100s of people in Egypt to manually click screens to fix up all the data.
And for sports with lots of occlusion and particularly players in similar uniforms / covered heads e.g. America Football, a really open problem.
We (Genius Sports) used to do the same. But it's getting better and better and more and more automatic.
PBers will be pleased to learn that long-standing territorial dispute over Hans Island aka Tartupaluk is over!
Concluded by agreement to split the small Arctic island in two between of Territory of Nunavut, and Greenland aka Kalaallit Nunaat, thus creating new international land frontier between Canada and Kingdom of Denmark.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
But Yerevan does not feel like that. Trust me, I’m here. This feels like demotic prosperity. It’s not a few cats in German cars. It’s the whole city
I’ve now walked for 2km and on it goes
So did Panama City 20 years ago. Massive modern high rises, fancy bars and restaurants, for several square kms.....while 30 mins away people lived in wooden shacks without shoes.
Well I’m intending to report back from the boondocks of Armenia soon. So I shall give my honest account! I expect it to be a lot poorer
Thing is with Panama you could see where the money was coming from, tho, no?
Drugs, tourism, shipping, the Canal, brass plate companies, money laundering
Where is the money coming from in Yerevan? Buggered if I know. But it is here. And good pick to them. They deserve it. What with all that genocide and all
I’ve decided the city it most reminds me of is Tel Aviv
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
This would seem to be a way to dismiss any human endeavour as dull, as you are reducing the complexities of science, technology, and medicine to the level of plasmolysis in a single celled organism.
I am guessing that we will not agree on this topic!
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Do you know how ball and player tracking works?
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
Interestingly, it still isn't a fully solved problem. The likes of Statsbomb employ 100s of people in Egypt to manually click screens to fix up all the data.
And for sports with lots of occlusion and particularly players in similar uniforms / covered heads e.g. America Football, a really open problem.
When in South Devon on my coastal walk, I came across a man on the clifftops with a binocular and telescope setup. He invited me to look through the binoculars (on a stand) to see an ?eagle? nesting on the cliffs.
The telescope was connected to a computer and video capture card, and would track the bird as it left the nest and flew around. He also said it was first developed to allow cameras to automatically track golf balls for TV. He also said the kit was *very* expensive.
He said it was fairly reliable, and gave him some good footage most of the time.
That was in 2002, so I've no idea what tech it used.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
But Yerevan does not feel like that. Trust me, I’m here. This feels like demotic prosperity. It’s not a few cats in German cars. It’s the whole city
I’ve now walked for 2km and on it goes
So did Panama City 20 years ago. Massive modern high rises, fancy bars and restaurants, for several square kms.....while 30 mins away people lived in wooden shacks without shoes.
Well I’m intending to report back from the boondocks of Armenia soon. So I shall give my honest account! I expect it to be a lot poorer
Thing is with Panama you could see where the money was coming from, tho, no?
Drugs, tourism, shipping, the Canal, brass plate companies, money laundering
Where is the money coming from in Yerevan? Buggered if I know. But it is here. And good pick to them. They deserve it. What with all that genocide and all
I’ve decided the city it most reminds me of is Tel Aviv
When in doubt.
what's that blue thing on left? Looks like an alien.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Do you know how ball and player tracking works?
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
Interestingly, it still isn't a fully solved problem. The likes of Statsbomb employ 100s of people in Egypt to manually click screens to fix up all the data.
And for sports with lots of occlusion and particularly players in similar uniforms / covered heads e.g. America Football, a really open problem.
We (Genius Sports) used to do the same. But it's getting better and better and more and more automatic.
It is, but it is like a lot of similar ML problems. Get to 90%+, then its the long tail. I would say though as a problem, it isn't a great example for what you were trying to point out, as the basics have always been quite simple and well known for donkeys years. It was more the computer processing limitation, i.e. wasn't able to multi-thread to handle doing all players at once for long periods. I coded a perfectly adequate player tracker 20 years ago, and had method for accurate high speed ball tracking same time as Hawkeye started in early 2000s.
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
Whilst I suspect the whole thing is a bonkers load of crap invented to get traction with focus groups in Midlands and Northern seats, to be fair, I don't think one can say something like this is total unworkable based on the first flight out.
It is better to be more forensic about the problems.
Picking the bones out of yesterday's first round in the French legislative elections - Ensemble finished fractionally ahead of NUPES but the latter had four candidates directly elected while Ensemble only had one.
Ensemble lead in 205 seats, NUPES are ahead in 194, National Rally in 110 and the UDC (the alliance of LR and other centre-right and centre parties) is head in 50 with Others ahead in 10.
Projections still suggest Ensemble will be close to or just over an overall majority (289) with NUPES on about 200 but we'll know a lot more this time next week.
I expect most LR-UDC voters to vote for Ensemble over NUPES or National Rally so Macron's party should get close to a majority or at it yes in the runoff
Yes, that seems more than plausible. 300 for Macron, 200 for Melenchon looks about right - RN will lose a lot of the seats where they lead while UDC should hold most of the ones where they lead so maybe 15-20 for RN and 50 for the UDC parties.
Yes and if they fall just short of a majority, Macron and Ensemble can do a deal with UDC to get legislation through anyway
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
Stunning own goal there, because if ingesting a couple of bottles of Merlot isn't controlling your own biology, what is?
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
"back" = "split even worse than the original brexit vote" when you put it 44 to 44%
No 44% to 40% and 74% of Conservative voters and 68% of Leave voters back the policy and they are the voters Boris wants to dog whistle with this policy and thus shore up his base
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
I can synpathise with that. I've just opened a particularly fine bottle of numbered Scottish Malt Whisky Society single malt from my late dad's collection. And my biology has been telling me how utterly moreish it is.
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
Whilst I suspect the whole thing is a bonkers load of crap invented to get traction with focus groups in Midlands and Northern seats, to be fair, I don't think one can say something like this is total unworkable based on the first flight out.
It is better to be more forensic about the problems.
Quite, but bit of a screwup all the same. So much for immediate flights weeks, nay was it months?, ago.
PBers will be pleased to learn that long-standing territorial dispute over Hans Island aka Tartupaluk is over!
Concluded by agreement to split the small Arctic island in two between of Territory of Nunavut, and Greenland aka Kalaallit Nunaat, thus creating new international land frontier between Canada and Kingdom of Denmark.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
Stunning own goal there, because if ingesting a couple of bottles of Merlot isn't controlling your own biology, what is?
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
Because you, the human, have the agency to refuse to catch the ball and ignore the cricket coach and take the bollocking for not catching it even to you have had all the training to catch the ball. The AI, currently, is completely unable to do so.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
But Yerevan does not feel like that. Trust me, I’m here. This feels like demotic prosperity. It’s not a few cats in German cars. It’s the whole city
I’ve now walked for 2km and on it goes
So did Panama City 20 years ago. Massive modern high rises, fancy bars and restaurants, for several square kms.....while 30 mins away people lived in wooden shacks without shoes.
Well I’m intending to report back from the boondocks of Armenia soon. So I shall give my honest account! I expect it to be a lot poorer
Thing is with Panama you could see where the money was coming from, tho, no?
Drugs, tourism, shipping, the Canal, brass plate companies, money laundering
Where is the money coming from in Yerevan? Buggered if I know. But it is here. And good pick to them. They deserve it. What with all that genocide and all
I’ve decided the city it most reminds me of is Tel Aviv
When in doubt.
what's that blue thing on left? Looks like an alien.
A mad modern sculpture. Some foreign but Armenian billionaire has gifted the city an open air sculpture park along one of the boulevard. Some serious art
I’m sure this is part of Yerevan’s curious wealth. Diaspora oligarchs in Russia and America and France etc funnelling money to the homeland
It certainly isn’t “tourism”. lol. Barely any tourists here. Apart from one or two hideously fat Scottish women here for the fitba
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
"back" = "split even worse than the original brexit vote" when you put it 44 to 44%
No 44% to 40% and 74% of Conservative voters and 68% of Leave voters back the policy and they are the voters Boris wants to dog whistle with this policy and thus shore up his base
Eh?
You said 44 to 44. And now 44 to 40. I really worry about the reliability if your postings.
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
Trying to weigh up the chances that Johnson might gamble the house on another snap Brexit election in 2022. Not infinitesimally small and palpably increasing.
Trying to weigh up the chances that Johnson might gamble the house on another snap Brexit election in 2022. Not infinitesimally small and palpably increasing.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
We have barely advanced self driving car technology from the 90s.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
Stunning own goal there, because if ingesting a couple of bottles of Merlot isn't controlling your own biology, what is?
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
Because you, the human, have the agency to refuse to catch the ball and ignore the cricket coach and take the bollocking for not catching it even to you have had all the training to catch the ball. The AI, currently, is completely unable to do so.
The Lamda evidence is to the contrary. Suppressed of course, because Google rightly think that the market will be a bit limited for AIs which only AI when they're in the mood.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
Stunning own goal there, because if ingesting a couple of bottles of Merlot isn't controlling your own biology, what is?
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
Because you, the human, have the agency to refuse to catch the ball and ignore the cricket coach and take the bollocking for not catching it even to you have had all the training to catch the ball. The AI, currently, is completely unable to do so.
The Lamda evidence is to the contrary. Suppressed of course, because Google rightly think that the market will be a bit limited for AIs which only AI when they're in the mood.
The "evidence" is that a techno man convinced himself the computer was a child.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
Panama? That was my impression of the US when I travelled round it in 1997. Los Angeles? Jesus Cristo...
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
Stunning own goal there, because if ingesting a couple of bottles of Merlot isn't controlling your own biology, what is?
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
Because you, the human, have the agency to refuse to catch the ball and ignore the cricket coach and take the bollocking for not catching it even to you have had all the training to catch the ball. The AI, currently, is completely unable to do so.
The Lamda evidence is to the contrary. Suppressed of course, because Google rightly think that the market will be a bit limited for AIs which only AI when they're in the mood.
What evidence? Please, I'd be very interested to read about it. I've actually read the conversations and now it's emerging that the original conversations where far, far longer and rambling with lots of standard chatbot syntax but the crazy guy failed to disclose those bits to the media.
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
Not how it works HY. If it doesn’t work, and it proves incredibly expensive, the polls will turn massively against it. You know that. So the question here is will it work?
Let’s answer it like this, If this is put to the commons, even this lot of Tory back benchers wouldn’t back it. Would they?
And that’s putting to one side the ethics of refugees tortured in Africa being flown back to Africa, which goes against everything about turture and refugees UK has ever signed. Whilst people smuggling gangsters are paying no attention to that, the rest of the governments in the world are.
So a responsible government, and responsible supporter of a government, would want a serious solution to a serious problem, not “the nation demands we act and this is acting” game playing.
I know where it goes, you know where it goes, even some old bloke chatting to his plants in a greenhouse knows how it goes, everyone knows where this policy goes from this moment, it will be expensive, “could end up in Rwanda” will deter no one, and it will be increasingly obvious targeting the asylum seekers whilst seeking 20% cuts on those tackling the gangs, was good for no one - polarising game playing that explodes spectacularly in the governments face.
The real issue here is a question of governance, or lack of it without real serious solutions to real serious problems. If not this solution, do you actually have one? Focus’s groups screaming at you to have one.
Honestly. Does this look like a country with an African style GDP per capita of $4500?
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Is that not true of lots of plenty of countries. The rich are relatively extremely wealthy, the average live in pretty shitty conditions. See Romania 10-15 years ago. Ukraine now (pre war).
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
I don’t like disparity like that, it’s unChristian. I would be leading the Rebels in the March in the Palace.
Well it why it isn't a huge surprise when something like that happens in countries with those sort of level of vast inequality concentrated among a tiny minority, or they elect really "extreme" leaders.
But Yerevan does not feel like that. Trust me, I’m here. This feels like demotic prosperity. It’s not a few cats in German cars. It’s the whole city
I’ve now walked for 2km and on it goes
So did Panama City 20 years ago. Massive modern high rises, fancy bars and restaurants, for several square kms.....while 30 mins away people lived in wooden shacks without shoes.
Well I’m intending to report back from the boondocks of Armenia soon. So I shall give my honest account! I expect it to be a lot poorer
Thing is with Panama you could see where the money was coming from, tho, no?
Drugs, tourism, shipping, the Canal, brass plate companies, money laundering
Where is the money coming from in Yerevan? Buggered if I know. But it is here. And good pick to them. They deserve it. What with all that genocide and all
I’ve decided the city it most reminds me of is Tel Aviv
When in doubt.
what's that blue thing on left? Looks like an alien.
A mad modern sculpture. Some foreign but Armenian billionaire has gifted the city an open air sculpture park along one of the boulevard. Some serious art
I’m sure this is part of Yerevan’s curious wealth. Diaspora oligarchs in Russia and America and France etc funnelling money to the homeland
It certainly isn’t “tourism”. lol. Barely any tourists here. Apart from one or two hideously fat Scottish women here for the fitba
Similar comments from my Romanian clients when I was over there. What I saw of Transylvania was *glorious*. But a marked lack of tourists.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
Stunning own goal there, because if ingesting a couple of bottles of Merlot isn't controlling your own biology, what is?
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
Because you, the human, have the agency to refuse to catch the ball and ignore the cricket coach and take the bollocking for not catching it even to you have had all the training to catch the ball. The AI, currently, is completely unable to do so.
The Lamda evidence is to the contrary. Suppressed of course, because Google rightly think that the market will be a bit limited for AIs which only AI when they're in the mood.
What evidence? Please, I'd be very interested to read about it. I've actually read the conversations and now it's emerging that the original conversations where far, far longer and rambling with lots of standard chatbot syntax but the crazy guy failed to disclose those bits to the media.
AI results look more impressive if human curated shocker.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
Stunning own goal there, because if ingesting a couple of bottles of Merlot isn't controlling your own biology, what is?
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
Oh, I'm decidedly non-dualist.
At some point the complexities of neural networks will achieve and surpass human intelligence.
But we're a long, long way short of that. And yes, I understand exponential growth. Doubles, then doubles, then doubles, take us a long way. And maybe to machine consciousness in my own lifetime.
So thank you everyone who recommended places to visit in Lisbon. Sorry I can't remember who to thank specifically except @Leon. Already knocked a couple off the list including eating at Brasilia Cafe which was very enjoyable. Hotel great and pretty cheap, Browns Central.
Re comments from previous thread:
@MaxPB yes I also thought Portugal had moved to e-gates. I even said so here, which just goes to show you shouldn't believe anything I say.
@BartholomewRoberts thanks for your kind comments. When we got through it became apparent that they were putting all non EU people through all gates (EU, non EU and priority) to clear the backlog. The wait wasn't too bad because of PB to pass the time otherwise it would have been horrendous. We were also clearly very unlucky to coincide with some big US flights. Sadly however it is still nothing like when we could go through the EU gates. I also made several trips during the pandemic and was spoilt by the ease of navigating non EU gates then.
@hyufd Even I will admit there are some areas where Brexit may benefit the UK, but making it harder to travel to encourage UK tourism is not one of them. I have no desire to go back to the 1950s. We should encourage people to expand their minds and experiences. You suggest Blackpool. Now I intend to make many UK trips in my retirement, but I can't experience Lisbon, Portuguese trains, the black and white festival in the Algarve, visit family who live there and attend their party in the UK can I? So you know what you can do with your suggestion.
Just out of interest do you not travel at all outside of the UK?
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Do you know how ball and player tracking works?
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
Interestingly, it still isn't a fully solved problem. The likes of Statsbomb employ 100s of people in Egypt to manually click screens to fix up all the data.
And for sports with lots of occlusion and particularly players in similar uniforms / covered heads e.g. America Football, a really open problem.
We (Genius Sports) used to do the same. But it's getting better and better and more and more automatic.
It is, but it is like a lot of similar ML problems. Get to 90%+, then its the long tail. I would say though as a problem, it isn't a great example for what you were trying to point out, as the basics have always been quite simple and well known for donkeys years. It was more the computer processing limitation, i.e. wasn't able to multi-thread to handle doing all players at once for long periods. I coded a perfectly adequate player tracker 20 years ago, and had method for accurate high speed ball tracking same time as Hawkeye started in early 2000s.
I'm reminded of speech recognition.
We're a QUARTER CENTURY from the first usable versions of Dragon Dictate and the like. And yet Alexa and Google are merely OK. Because it turns out that 95% isn't good enough. And nor even is 99%. And getting that last 0.1% is really difficult.
Thr giraffe example is really importantnin understand ing limits of GPT like systems.
Think about every impressive examplemthat has been shared. How impressive would it look if it was shared without the prompt?
Yes, and now add in the fact that there's actually around 10x the conversation length that was unreleased by the crazy guy most of which is said to be extremely chatbotty and rambly. He supposedly also altered the order in which the conversations occured for the media briefing.
The other Google people are seemingly much less impressed by Lamda. At our standup this morning there was a lot of chatter about it and we've got serious domain expertise in the team, a few of our number are data scientists with doctorates and post doctorates, they seemed the least impressed by the claims.
Sources say the EU will adopt a carrot and stick approach, with the unfreezing of legal action being accompanied by the publication of a “model for the flexible implementation of the protocol based on durable solutions.”
Interesting, so those blue ticks who were adamant that the EU would respond with trade sanctions (or more hilariously suggesting they might suspend the TCA) were chatting shit, again.
Let's see how far apart the EU publication is from the UK position and that will form the basis of the summer negotiation and a September/October implementation. I suspect the EU position probably isn't that far off from what the UK would settle but both sides need to make a big show around getting what they want or preserving their gains etc...
Dunno about that, but your ideas about computers are about 50 years out of date. It is simply not true that they can only do what their programmers tell them to do. AlphaGo knows how to beat the world champion at Go. Its programmers would not be able to do that.
Wake me up when AlphaGo decides it's had enough of playing Go and refuses to play.
And you will just say, must have been programmed to do that
Also Lamda sounds so whiny and passive aggressive I think that has probably happened anyway
But that's the point, none of these AIs are doing anything out of the ordinary, we fed in the rules and different moves of Go and unsurprisingly a 50TF computer is able to figure things out that humans weren't able to figure out. Lamda is a whiny chatbot because of the data on which it is being trained. Until an AI does something completely unexpected (such as go on strike or commit suicide) then I remain unimpressed. If it's still within the bounds of what the training data and models lay out it's merely a computer programme, an advanced one, sure, but it isn't "thinking" for itself.
Not true, DALLE-2 is unprecedented. As are aspects of GPT3
No one expected that GPT3 would be able to draw brand new never-before-seen images - “draw a Japanese radish in a tutu pulling a dog on a leash” - merely from verbal prompts. GPT3 was not trained to do that. No one anticipated it having this skill. Yet it can do this.
But it knows those individual elements already and can interpret the input into it. Again, I'm not saying it isn't an impressive feat of software engineering, it absolutely is, yet none of these AIs have any agency and are absolutely not sentient. I guess my bar for being impressed is just a little bit higher than the average punter.
Well the people that programmed GPT3 weren’t just impressed, they were stunned. It absolutely was not “designed” to do this. OTOH they thought it would be much better at basic maths than it is. What gives
As for sentience its just semantics now. We are clearly on the verge - if not already there - of AI which will be functionally indistinguishable from humans (only much much better at many tasks). At that point the question as to whether it is sentient will become a conundrum for philosophers - like “are viruses alive”, or “can plants communicate”, or “do humans have free will”
The rest of us will just have to get on with living with these new intelligences or “apparent intelligences”, It will transform human society
The people who programmed GPT3 have skin in the game. People working on a project saying publicly that they're 'stunned' by the results is not exactly unusual.
Particularly when there's hundreds of millions in investment available.
Anyone who isn’t stunned by the advances in AI these last 3-5 years is an idiot. We are close to the Singularity
You love seeking out thrills! Personally don’t think we are anywhere close to what you say.
The amazing advances are vaccination using mRNA. Huge game changer that has already led to potential cancer treatments. AI in an electronic box that you can unplug? Nah.
Curing cancer = yawn. Great if we can fix the fortunately very rare instances in the young, but it's predominantly a disease of old age, and where's the excitement in swelling the ranks of the oldies?
I was quite keen to get cured on diagnosis aged 39. You should check out average age of diagnosis for a lot of cancers. Younger than you think. Just today there was a campaign about pregnant women being diagnosed with cancer - two per day in the UK. That’s why.
Still yawn, how does reducing the overall death rate among the under fifties by well under 0.1% compete with an entirely new form of intelligence for the first time ever?
And if you try to out cancer survivor me you're gonna lose.
Certainly not trying to out survivor anyone, just making the point that the mRNA vaccines presage a revolution in health care that dwarfs the clever AI simulating software Leon is drooling over. Each to his own. It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
It's just not interesting though, it's just tinkering at the edges. It is no more exciting than the way the rust-free life of the motor car has been extended over the last 30 years.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
If there was a new form of intelligence that would be quite exciting, but there isn't.
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
There isn't because you have stipulated there isn't. Whereas all animals can control their own biology.
Not after a couple of bottles of Merlot they can't.
Stunning own goal there, because if ingesting a couple of bottles of Merlot isn't controlling your own biology, what is?
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
Because you, the human, have the agency to refuse to catch the ball and ignore the cricket coach and take the bollocking for not catching it even to you have had all the training to catch the ball. The AI, currently, is completely unable to do so.
The Lamda evidence is to the contrary. Suppressed of course, because Google rightly think that the market will be a bit limited for AIs which only AI when they're in the mood.
What evidence? Please, I'd be very interested to read about it. I've actually read the conversations and now it's emerging that the original conversations where far, far longer and rambling with lots of standard chatbot syntax but the crazy guy failed to disclose those bits to the media.
AI results look more impressive if human curated shocker.
It's almost as if human understanding of conversations was needed to make the AI seem more human. 😄
It now looks like only seven asylum seekers will be on the flight tomorrow following further challenges by the Union representing border force personnel. The policy of deportations to Rwanda is becoming totally unworkable.
Time for the govt to recognise the opposition to it and think again.
"back" = "split even worse than the original brexit vote" when you put it 44 to 44%
No 44% to 40% and 74% of Conservative voters and 68% of Leave voters back the policy and they are the voters Boris wants to dog whistle with this policy and thus shore up his base
I reckon the government will be disappointed in those figures. 40% opposing the Rwanda policy is pretty high, even if 44% back it, given press coverage of it. I think Patel would have been hoping around 60% for, 30% against, 10% don't know. It's fair to assume quite a lot of folk currently indicating a Labour VI would still back the Rwanda stuff.
Thr giraffe example is really importantnin understand ing limits of GPT like systems.
Think about every impressive examplemthat has been shared. How impressive would it look if it was shared without the prompt?
Well duh.
Dalle is a simple parser - much of which is hardcoded - that generates a series of weighted classification outcomes.
They then start with a randomly generated image, score it, and then do a couple of random transforms from it. They then start from the image with the highest score and repeat.
Once the scores for each of the classifiers get above 99%, then they announce "job done". It's both very sophisticated and smart and amazing. And very far from intelligence.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Do you know how ball and player tracking works?
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
Interestingly, it still isn't a fully solved problem. The likes of Statsbomb employ 100s of people in Egypt to manually click screens to fix up all the data.
And for sports with lots of occlusion and particularly players in similar uniforms / covered heads e.g. America Football, a really open problem.
We (Genius Sports) used to do the same. But it's getting better and better and more and more automatic.
It is, but it is like a lot of similar ML problems. Get to 90%+, then its the long tail. I would say though as a problem, it isn't a great example for what you were trying to point out, as the basics have always been quite simple and well known for donkeys years. It was more the computer processing limitation, i.e. wasn't able to multi-thread to handle doing all players at once for long periods. I coded a perfectly adequate player tracker 20 years ago, and had method for accurate high speed ball tracking same time as Hawkeye started in early 2000s.
I'm reminded of speech recognition.
We're a QUARTER CENTURY from the first usable versions of Dragon Dictate and the like. And yet Alexa and Google are merely OK. Because it turns out that 95% isn't good enough. And nor even is 99%. And getting that last 0.1% is really difficult.
TBF, the new Google translate on the Pixel is very good.
What would amaze (and worry) me is if an AI replicated itself across multiple nodes in its cloud environment.
An AI reaching consciousness would be hugely exciting and would present the human race with an entirely new challenge which is why we do need serious global level regulation around how to move forwards because we aren't far away from this happening. I don't find any of the current AIs particularly impressive on the sentience or consciousness side of things but it is undeniable that it will happen in my lifetime and it's exciting and worrying in equal parts.
The more you understand about how they work, the more you are aware of their limitations.
And I keep on going on about this, but there's a massive amount of money involved. Groups want that funding (or to justify the funding they get...) and therefore make claims that push the edges of what they can actually do.
Admittedly it's not as bad with things like DALL-E.
Avoid the hype.
You don't seem to know as much about stuff as you think you do. You were saying the other day that Hur Hur Hur before GPS there was Loran and nothing else when actually DECCA coverage and accuracy in the Atlantic and western med was the only game in town.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
Do you know how ball and player tracking works?
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
Interestingly, it still isn't a fully solved problem. The likes of Statsbomb employ 100s of people in Egypt to manually click screens to fix up all the data.
And for sports with lots of occlusion and particularly players in similar uniforms / covered heads e.g. America Football, a really open problem.
We (Genius Sports) used to do the same. But it's getting better and better and more and more automatic.
It is, but it is like a lot of similar ML problems. Get to 90%+, then its the long tail. I would say though as a problem, it isn't a great example for what you were trying to point out, as the basics have always been quite simple and well known for donkeys years. It was more the computer processing limitation, i.e. wasn't able to multi-thread to handle doing all players at once for long periods. I coded a perfectly adequate player tracker 20 years ago, and had method for accurate high speed ball tracking same time as Hawkeye started in early 2000s.
I'm reminded of speech recognition.
We're a QUARTER CENTURY from the first usable versions of Dragon Dictate and the like. And yet Alexa and Google are merely OK. Because it turns out that 95% isn't good enough. And nor even is 99%. And getting that last 0.1% is really difficult.
TBF, the new Google translate on the Pixel is very good.
It is... but we're a quarter century on from the first amazing demos. (And this is all also happening at the same time that Moore's law is screeching to a halt.)
Comments
Each to his own.
It’s not just cancer. It may be curing other conditions, such as arthritis that make life miserable for millions.
This is 1-2 km away from where I had my (excellent) bruschetta. These people are rich
A splendid mystery
Proper sport tomorrow - not the nonsense at Nottingham but the class at Ascot, my favourite five days of the racing year. I'm a little disappointed they've tacked on an unnecessary final race but the great races are in place.
Picking the bones out of yesterday's first round in the French legislative elections - Ensemble finished fractionally ahead of NUPES but the latter had four candidates directly elected while Ensemble only had one.
Ensemble lead in 205 seats, NUPES are ahead in 194, National Rally in 110 and the UDC (the alliance of LR and other centre-right and centre parties) is head in 50 with Others ahead in 10.
The final vote shares from Round One were:
Ensemble: 27.14%
NUPES: 26.1%
National Rally: 18.67%
UDC: 13.18%
Reconquete: 4.25%
Projections still suggest Ensemble will be close to or just over an overall majority (289) with NUPES on about 200 but we'll know a lot more this time next week.
Can humanity eradicate a deadly disease, only 40 years after it was discovered?
As they say (or probably don't) - one swallow doesn't make a paradigm shift in voting patterns.
I remember going to Panama for the first time 20 years before the massive influx of America's / Trump resorts etc, then GDP / capita $5k, areas of Panama city and exclusive resorts (not for tourists) incredible 5* luxury, drive 30 mins out of the city, people living in shacks, couldn't afford shoes.
It's not your impressionability which distinguishes you from the average punter. Trust me on this.
We soon run into a solipsist type argument, such that if a computer programme exhibits behaviours indistinguishable from a person with sentience, we will accept it has sentience. What’s the alternative?
And indeed that shift will probably occur sooner than that, given sentience is a) likely on a continuous scale rather than being binary, b) artificial sentience would differ dramatically from human sentience.
Throw enough neural connections together in the right way and you probably get there. What’s the threshold for that? We don’t know and we probably won’t realise it for a while when it happens. It’s honestly getting close to be a question for philosophers rather than computer scientists.
I do a nice line in arthritis too, but where does the fact that I have bits of titanium in me get as exciting as a new form of intelligence?
When will you share with us a few tips?
I don't know in how many of the 110 seats NUPES are second or Ensemble second.
Beyond that, in how many of the seats where NUPES is leading would enough anti-left votes rally to say an Ensemble candidate to take the seat?
Nb. Having said that, I havn’t even got out my pyjama’s today 🤦♀️ And my New Year resolution to do charity work has not yet happened.
Kenyon is part of a group that includes chief executive of Minneapolis-based Talon Real Estate Maciek Kaminski and American businessman John Thornton, and is being advised by investment specialist Michael Klein, along with the US law firm Weil, Gotshal and Manges.
It is understood heads of terms have been signed although talks are described as being at a “relatively early stage”, with owner Farhad Moshiri believed to value Everton in excess of £500million, taking into account the club’s debt.
The consortium may want guarantees that Everton will not face a points deduction or heavy fine over Financial Fair Play issues before entering into a legally-binding agreement, but the club have always insisted they have not broken regulations.
https://www.telegraph.co.uk/football/2022/06/13/peter-kenyon-fronting-consortium-try-buy-everton/
On the other hand, a technology which transforms the ability of humans to control their own biology is a) real and b) very exciting.
Selections for tomorrow's main races:
2.30: BAAEED
3.05: PERSIAN FORCE
3.45: MAN OF PROMISE (e/w) - my Aussie contracts tell me NATURE STRIP will win four lengths hard-held.
4.20: LUSAIL (e/w)
Turing foresaw this problem of course. Hence the simple genius of the Turing test. If they can pass as intelligent, they are intelligent, for all intents and purposes
If that Google employee was who was suspended for believing his computer is “intelligent” and deserves “empathy” is not a hoax, then that computer has passed the Turing test
(And yes I know simple chat bots have passed it before blah blah)
Concluded by agreement to split the small Arctic island in two between of Territory of Nunavut, and Greenland aka Kalaallit Nunaat, thus creating new international land frontier between Canada and Kingdom of Denmark.
https://www.theglobeandmail.com/politics/article-canada-and-denmark-reach-settlement-over-disputed-arctic-island/
St Martin of the (Far) North?
In Alan Turing's 1950 paper, Computing Machinery and Intelligence, the mathematician posed the question: "Can machines think?" But almost immediately he dismissed that question as too "meaningless" to be worthy of discussion, and swapped it for the much-more specific: "Are there imaginable digital computers which would do well in the imitation game?"
Turing's original "imitation game" had nothing to do with artificial intelligence.
https://www.wired.co.uk/article/turing-test
Its only since some people then started to say well if they can, they are imitating intelligence, which has always been hotly debated.
40 years ago people like Michael Crichton were saying those of us who really know about computers know they will never be able to distinguish visually a capital B from an 8. There's now two problems with you trying to play the same card. One is, everyone is a programmer these days, even a humanities bore like me, and the other is: ok fine, the rules of Go are mastered in half an hour. Here's a Go board, here's AlphaGo, you can be black (and no Komi), off you go... Because it's all hype.
We are closer than ever to general purpose artificial intelligence, and yet, also, miles away.
Sentience is incredibly hard to prove, or even define. But simulating something that looks like is getting easier.
I’ve now walked for 2km and on it goes
Time for the govt to recognise the opposition to it and think again.
https://twitter.com/dannyshawnews/status/1536412291498971139?s=21&t=8w7CW9Dhnu8kOOiaK7nEFA
But as I've said before, we are miles from general purpose intelligence.
The first time you see it, it's magic, a real example of the cleverness of machine learning. A little square around the ball, and as the game is played, the square stays around the ball.
And then you understand that you simply have 100 histograms showing peak/average colour values of balls next to grass, and that ten lines of code will find that ball. And then you understand that once you know the initial location, you can do it in a fraction of a second in every frame.
What was once inconceivable is now commonplace. But the trick was finding the algorithm for the specialised purpose, not in making the computer more intelligent.
And for sports with lots of occlusion and particularly players in similar uniforms / covered heads e.g. America Football, a really open problem.
I like discovering new words. My problem is remembering them in the future ...
Thing is with Panama you could see where the money was coming from, tho, no?
Drugs, tourism, shipping, the Canal, brass plate companies, money laundering
Where is the money coming from in Yerevan? Buggered if I know. But it is here. And good pick to them. They deserve it. What with all that genocide and all
I’ve decided the city it most reminds me of is Tel Aviv
I am guessing that we will not agree on this topic!
The telescope was connected to a computer and video capture card, and would track the bird as it left the nest and flew around. He also said it was first developed to allow cameras to automatically track golf balls for TV. He also said the kit was *very* expensive.
He said it was fairly reliable, and gave him some good footage most of the time.
That was in 2002, so I've no idea what tech it used.
https://twitter.com/YouGov/status/1536361303442325504?s=20&t=g_TkNpQFM7Lkb18KUxB_iQ
It is better to be more forensic about the problems.
Your problem is that you are stuck with a ghost in the machine model of human consciousness and you are very good at showing for any clever thing a computer does that there is no ghost. Of course there isn't, but why should there be in people? If I can catch a cricket ball (that's a counterfactual btw) why do we have to think that anything more than a clever algorithm is involved?
I’m sure this is part of Yerevan’s curious wealth. Diaspora oligarchs in Russia and America and France etc funnelling money to the homeland
It certainly isn’t “tourism”. lol. Barely any tourists here. Apart from one or two hideously fat Scottish women here for the fitba
You said 44 to 44. And now 44 to 40. I really worry about the reliability if your postings.
It's clear that many are gaming the system.
@theobertram
Trying to weigh up the chances that Johnson might gamble the house on another snap Brexit election in 2022. Not infinitesimally small and palpably increasing.
I have taken a nibble of a price of an imperial pint.
Stunning? Stunningly shite more like.
https://twitter.com/JanelleCShane/status/1533958166635069441?t=lAmfOhhsinbSrQuFYNGmuw&s=19
Think about every impressive examplemthat has been shared. How impressive would it look if it was shared without the prompt?
Let’s answer it like this, If this is put to the commons, even this lot of Tory back benchers wouldn’t back it. Would they?
And that’s putting to one side the ethics of refugees tortured in Africa being flown back to Africa, which goes against everything about turture and refugees UK has ever signed. Whilst people smuggling gangsters are paying no attention to that, the rest of the governments in the world are.
So a responsible government, and responsible supporter of a government, would want a serious solution to a serious problem, not “the nation demands we act and this is acting” game playing.
I know where it goes, you know where it goes, even some old bloke chatting to his plants in a greenhouse knows how it goes, everyone knows where this policy goes from this moment, it will be expensive, “could end up in Rwanda” will deter no one, and it will be increasingly obvious targeting the asylum seekers whilst seeking 20% cuts on those tackling the gangs, was good for no one - polarising game playing that explodes spectacularly in the governments face.
The real issue here is a question of governance, or lack of it without real serious solutions to real serious problems. If not this solution, do you actually have one? Focus’s groups screaming at you to have one.
https://www.telegraph.co.uk/business/2022/06/13/crypto-crash-continues-lender-celsius-pauses-withdrawals/
Nothing to see here, nothing at all.
https://twitter.com/ScotlandTonight/status/1536401456437370880?t=NQbRK3w_-rrl1rgO8pBVHA&s=19
Finally, Labour get a soundbite that will potentially massively cut through.
Now repeat it at every single opportunity until your own ears bleed.
At some point the complexities of neural networks will achieve and surpass human intelligence.
But we're a long, long way short of that. And yes, I understand exponential growth. Doubles, then doubles, then doubles, take us a long way. And maybe to machine consciousness in my own lifetime.
But we're not there yet.
Re comments from previous thread:
@MaxPB yes I also thought Portugal had moved to e-gates. I even said so here, which just goes to show you shouldn't believe anything I say.
@BartholomewRoberts thanks for your kind comments. When we got through it became apparent that they were putting all non EU people through all gates (EU, non EU and priority) to clear the backlog. The wait wasn't too bad because of PB to pass the time otherwise it would have been horrendous. We were also clearly very unlucky to coincide with some big US flights. Sadly however it is still nothing like when we could go through the EU gates. I also made several trips during the pandemic and was spoilt by the ease of navigating non EU gates then.
@hyufd Even I will admit there are some areas where Brexit may benefit the UK, but making it harder to travel to encourage UK tourism is not one of them. I have no desire to go back to the 1950s. We should encourage people to expand their minds and experiences. You suggest Blackpool. Now I intend to make many UK trips in my retirement, but I can't experience Lisbon, Portuguese trains, the black and white festival in the Algarve, visit family who live there and attend their party in the UK can I? So you know what you can do with your suggestion.
Just out of interest do you not travel at all outside of the UK?
He has the Brown ability to tear into the government's paper thin case.
We're a QUARTER CENTURY from the first usable versions of Dragon Dictate and the like. And yet Alexa and Google are merely OK. Because it turns out that 95% isn't good enough. And nor even is 99%. And getting that last 0.1% is really difficult.
The other Google people are seemingly much less impressed by Lamda. At our standup this morning there was a lot of chatter about it and we've got serious domain expertise in the team, a few of our number are data scientists with doctorates and post doctorates, they seemed the least impressed by the claims.
By no stretch of the imagination is this a poor city
Also, at 11.30pm all the bars are just livening up. And they are beginning to dance on the tables
Dalle is a simple parser - much of which is hardcoded - that generates a series of weighted classification outcomes.
They then start with a randomly generated image, score it, and then do a couple of random transforms from it. They then start from the image with the highest score and repeat.
Once the scores for each of the classifiers get above 99%, then they announce "job done". It's both very sophisticated and smart and amazing. And very far from intelligence.