Time for a new party that represents Muslims adequately. We will not win overall but could easily win parliamentary seats. We are currently in active discussions to get this done. Muslim Labour MPs have let the community down. Change is needed.
they will want England to be like Palestine next
Full of dead Brown People?
Not really.
You should stick to your area of expertise Malc.!
What drove you off the edge BJ, you were normal on here once.
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts, which it gets wrong and I could write it myself very quickly) then to say python (which GPT3 could do) and result of this is still so so.
Even if you give it the whole page with all the details of what the various symbols mean, unless its trivial, its not great at it. And if its trivial, a human coder can do it just as fast anyway.
Again, will this get better and better, absolutely.
Tbh I know fuck all about coding, and care less (other than it’s huge secondary impact on our lives, of course)
What I do know and care about is the arts, writing, design, advertising, drawing, thinking, all the interesting stuff the mind does - and if GPT5 is as better at writing (etc) than GPT4 - as GPT4 was better compared to GPT3 - then holy fuck
Yeah, AI is great at the fluffy stuff and will only get better. But it has major difficulties with areas that require accuracy and actual understanding, like maths and coding. It'll have a go at them, but its efforts are a bit cargo-culty, in that they look the part but are often missing some fundamental aspect.
Space, the final frontier. Also the solution to pollution for a little while, We really can dump stuff into the sun for centuries without it getting pissed off. We'll have to move on from that though.
I always wondered why we never tried that with nuclear waste.
Earth surface to Earth orbit Much as we like to think it's all sorted, a certain percentage of rockets still go bang. Nobody wants to have nuclear debris scattered across their country. So nobody wants to do it.
Earth orbit to Sun An object in order around the Sun has a lot of energy pointing at right-angles to the Sun: it's only the Sun's gravity that stops it speeding off. The greater the change in velocity or direction[1], the more fuel you need. If you want to send it straight into the Sun you need a LOT of fuel. If you want to send it spiralling in, that's cheaper but it takes longer. It may genuinely be easier to send it outwards to Mars/Jupiter, slingshot around it and head back into the Sun.
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts, which it gets wrong and I could write it myself very quickly) then to say python (which GPT3 could do) and result of this is still so so.
Even if you give it the whole page with all the details of what the various symbols mean, unless its trivial, its not great at it. And if its trivial, a human coder can do it just as fast anyway.
Again, will this get better and better, absolutely.
Tbh I know fuck all about coding, and care less (other than it’s huge secondary impact on our lives, of course)
What I do know and care about is the arts, writing, design, advertising, drawing, thinking, all the interesting stuff the mind does - and if GPT5 is as better at writing (etc) than GPT4 - as GPT4 was better compared to GPT3 - then holy fuck
Yeah, AI is great at the fluffy stuff and will only get better. But it has major difficulties with areas that require accuracy and actual understanding, like maths and coding. It'll have a go at them, but its efforts are a bit cargo-culty, in that they look the part but are often missing some fundamental aspect.
Asking for actual understanding is a simple category error. Its like asking a tree to swim and then scoffing when it can’t
"Joanna Cherry KC @joannaccherry Those posting snide comments about Ash Regan defection to Alba should stop & reflect on whether they have played a part in splitting the independence movement. The world & our country face such huge problems I wd like to see reconciliation not more division 2:30 PM · Oct 28, 2023"
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts) then to say python and still so so.
It just takes one of these machines to have the ability to self improve - then KABOOM
It doesn't even "understand" basic maths yet, let alone know how to improve upon that. I had it the other day arguing black is white over the incorrect details of a tensor, which is trivial maths.
This is the fundamental issue with the idea of LLM being predicted on probabilistic selection of next token.
Arguably it doesn't "understand" anything because it's a goldfish in a bowl with a 5 second memory that happens to be born with an encyclopedia inside its head.
Give it long term memory, and it starts to look shockingly like how human minds work.
An LLM with long term memory accessible the way humans can remember something from 20 years ago at a pinch, starts to look shockingly like us. Give it the ability to improve on its own code and it starts to look very much like us. The human mind is nothing special.
We may be 5-10 years away from this, or some have speculated, Google or OpenAI may be there already but holding back on releasing it for fear of what it will do. But it is coming.
I think need to be careful what I mean by understanding. No its doesn't understand anything in the human sense, but I would argue it has learned very decent set of rules for natural language and that the probabilistic approach then allows for this nice ability to produce psuedo-creative outputs.
My experience working with it, this training approach doesn't seem to have worked very well for ascertaining the rules of maths or coding of maths. There is a reason they hooked it to mathematica, because without, it is actually really bad. Can this be improved, sure, but is the probabilistic LLM approach to trying to model hard science like maths the right approach, I am not convinced. Could there by something similar or better, sure. But we aren't there yet. The human created Mathematica is still vastly superior, and everybody in STEM isn't saying well that all the jobs gone because of Wolfram Alpha / Mathematica (which has been about for what 35 years), rather its a really useful tool.
Time for a new party that represents Muslims adequately. We will not win overall but could easily win parliamentary seats. We are currently in active discussions to get this done. Muslim Labour MPs have let the community down. Change is needed.
they will want England to be like Palestine next
Full of dead Brown People?
Not really.
You should stick to your area of expertise Malc.!
What drove you off the edge BJ, you were normal on here once.
Wanting to stop Kids being slaughtered is normal 76% of people agree with me
Wanting the slaughter to continue is being "driven off the edge" and "Normal" No
The Israel Defense Force (IDF) and Israel Securities Authority (ISA) have released a joint statement in the last few minutes announcing the rescue of a female Israeli soldier taken hostage by Hamas.
Time for a new party that represents Muslims adequately. We will not win overall but could easily win parliamentary seats. We are currently in active discussions to get this done. Muslim Labour MPs have let the community down. Change is needed.
they will want England to be like Palestine next
Full of dead Brown People?
Not really.
You should stick to your area of expertise Malc.!
What drove you off the edge BJ, you were normal on here once.
Wanting to stop Kids being slaughtered is normal 76% of people agree with me
Wanting the slaughter to continue is being "driven off the edge" and "Normal" No
Listen to yourself mate
Yes, the polling on ceasefire is quite strongly in favour, even amongst Tory voters.
Not that either Israel or Hamas are likely to listen to anyone in this country.
Time for a new party that represents Muslims adequately. We will not win overall but could easily win parliamentary seats. We are currently in active discussions to get this done. Muslim Labour MPs have let the community down. Change is needed.
they will want England to be like Palestine next
Full of dead Brown People?
Not really.
You should stick to your area of expertise Malc.!
What drove you off the edge BJ, you were normal on here once.
Wanting to stop Kids being slaughtered is normal 76% of people agree with me
Wanting the slaughter to continue is being "driven off the edge" and "Normal" No
Listen to yourself mate
Yes, the polling on ceasefire is quite strongly in favour, even amongst Tory voters.
Not that either Israel or Hamas are likely to listen to anyone in this country.
Anna Soubry (yes, I know I know) was on Sky News yesterday doing the newspaper reviews and made a good point about the disagreements in Labour. She said on an issue like this, where there is no position you can take that isn't problematic in one way or another, and where people are agonising in their own minds about the moral dilemmas involved, it is perfectly reasonable for a party to be debating loudly. Indeed a good thing.
I think Starmer could make a virtue of this, particularly after the Bristow sacking. It's in some ways a matter of conscience because the debate isn't really over British policies per se. Let people voice their own views on this so long as they avoid Antisemitism or Islamophobia.
New Labour's conviction that everyone could enjoy the graduate premium is up there with Michael Gove's desire for every school to be above average.
If I remember correctly all these headlines are based upon an academic paper from a few months ago, which argued that it isn't simply the increase in graduates, it is that the number of graduate level / graduate paying jobs hasn't increased to keep in line with this. And that in comparison, the US have managed to do continue to expand their graduate level job creation, and the only in London has this kept apace (and no its not just since Brexit, its over the past 20 years).
Furthermore, the argument presented in the paper is that isn't just too many mickey mouse degrees, but again the US has plenty of those as well.
Now, I might say that its rather circular argument, you have lots of STEM graduates, you more likely to get high tech industry, which then spills over across the board. That is certainly what has happened in the US cities like San Francisco.
I think we all know for too long productivity has also been a massive issue in the UK.
White House Office of Science & Technology Policy @WHOSTP The Biden-Harris Administration is taking the most significant action any government has ever taken on AI. With today's executive order, we are advancing safe, secure, and trustworthy AI. Head to http://AI.gov to learn more.
Has the White House always had Biden-Harris as its moniker or is there an election coming up? I genuinely can't remember whether they've always given Kamala equal billing.
Time for a new party that represents Muslims adequately. We will not win overall but could easily win parliamentary seats. We are currently in active discussions to get this done. Muslim Labour MPs have let the community down. Change is needed.
they will want England to be like Palestine next
Full of dead Brown People?
Not really.
You should stick to your area of expertise Malc.!
What drove you off the edge BJ, you were normal on here once.
Wanting to stop Kids being slaughtered is normal 76% of people agree with me
Wanting the slaughter to continue is being "driven off the edge" and "Normal" No
Listen to yourself mate
Yes, the polling on ceasefire is quite strongly in favour, even amongst Tory voters.
Not that either Israel or Hamas are likely to listen to anyone in this country.
"Ceasefire" means both sides stop hostilities.
That means: - All hostages returned unharmed (clearly already impossible, so there's an immediate problem) - Hamas/Islamic Jihad/etc stop firing rockets at Israeli cities - Hamas leadership surrender and are deported to the Hague to face justice for the crimes against humanity they've committed
Once that's done, Israel will have no reason to continue the war anyway.
Of course the majority of people want a ceasefire. The issue is that the above isn't happening anytime soon, or ever.
New Labour's conviction that everyone could enjoy the graduate premium is up there with Michael Gove's desire for every school to be above average.
As the FT reporting this week showed, the US has managed to add graduates at the same rate as the UK while actually increasing the graduate pay premium.
"Black Met officers urged by colleagues to donate to fund for sacked PCs Exclusive: Black officers complain of collection for men who lied about the stop and search of black athletes"
Isn't this the sort of thing The Home Secretary should be criticising? Or is there some reason that she might not want to, that I can't think of right now...
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts) then to say python and still so so.
It just takes one of these machines to have the ability to self improve - then KABOOM
It doesn't even "understand" basic maths yet, let alone know how to improve upon that. I had it the other day arguing black is white over the incorrect details of a tensor, which is trivial maths.
This is the fundamental issue with the idea of LLM being predicted on probabilistic selection of next token.
Arguably it doesn't "understand" anything because it's a goldfish in a bowl with a 5 second memory that happens to be born with an encyclopedia inside its head.
Give it long term memory, and it starts to look shockingly like how human minds work.
An LLM with long term memory accessible the way humans can remember something from 20 years ago at a pinch, starts to look shockingly like us. Give it the ability to improve on its own code and it starts to look very much like us. The human mind is nothing special.
We may be 5-10 years away from this, or some have speculated, Google or OpenAI may be there already but holding back on releasing it for fear of what it will do. But it is coming.
I think need to be careful what I mean by understanding. No its doesn't understand anything in the human sense, but I would argue it has learned very decent set of rules for natural language and that the probabilistic approach then allows for this nice ability to produce psuedo-creative outputs.
My experience working with it, this training approach doesn't seem to have worked very well for ascertaining the rules of maths or coding of maths. There is a reason they hooked it to mathematica, because without, it is actually really bad. Can this be improved, sure, but is the probabilistic LLM approach to trying to model hard science like maths the right approach, I am not convinced. Could there by something similar or better, sure. But we aren't there yet. The human created Mathematica is still vastly superior, and everybody in STEM isn't saying well that all the jobs gone because of Wolfram Alpha / Mathematica (which has been about for what 35 years), rather its a really useful tool.
Yes, I agree with you entirely there. About 50% of my queries to GPT4 involve the Wolfram plugin - e.g. here are a bunch of figures, can you go away and calculate NPV on this project? With Wolfram it gets it right, without, it's a mess. So it has no intelligence or understanding in the human sense, yet.
The interesting question I keep coming back to is, what does intelligence, or as you call it, understanding, really look like, and how will we distinguish it from a really advanced inference model when we reach it?
Most of the people I know working in the AI space at the moment are of the opinion that humans are just really advanced LLMs with gigantic memories and processing power. I suspect this theory will be proven in the next few years - though it's probably harder to disprove.
I like to think back to Data from TNG. Is he really alive, or just a very very fast processor chip with memory access and a 24th century LLM? What makes human consciousness special or unique?
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
No apparent Palestine backlash for Labour yet, unless those two Green points are disillusioned Labour Left. LD have been very stable in recent polls across most pollsters. Feels like the Dutch salute has been done, the top remains up, people have made a settled choice. I'd expected a fall after the Mid-Beds result but nothing obvious.
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
Time for a new party that represents Muslims adequately. We will not win overall but could easily win parliamentary seats. We are currently in active discussions to get this done. Muslim Labour MPs have let the community down. Change is needed.
they will want England to be like Palestine next
Full of dead Brown People?
Not really.
You should stick to your area of expertise Malc.!
What drove you off the edge BJ, you were normal on here once.
Wanting to stop Kids being slaughtered is normal 76% of people agree with me
Wanting the slaughter to continue is being "driven off the edge" and "Normal" No
Listen to yourself mate
Yes, the polling on ceasefire is quite strongly in favour, even amongst Tory voters.
Not that either Israel or Hamas are likely to listen to anyone in this country.
I'm not sure even as simple a concept as ceasefire has a universally accepted interpretation in this instance, with some expecting it to involve return of hostages and others not.
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
White House Office of Science & Technology Policy @WHOSTP The Biden-Harris Administration is taking the most significant action any government has ever taken on AI. With today's executive order, we are advancing safe, secure, and trustworthy AI. Head to http://AI.gov to learn more.
Has the White House always had Biden-Harris as its moniker or is there an election coming up? I genuinely can't remember whether they've always given Kamala equal billing.
Answer is door #2 = there's an election coming up.
Same as with all but one administration in my lifetime.
During the Cheney-Bush administration the head honcho graciously gave quasi-equal billing to HIS sidekick, W.
Time for a new party that represents Muslims adequately. We will not win overall but could easily win parliamentary seats. We are currently in active discussions to get this done. Muslim Labour MPs have let the community down. Change is needed.
they will want England to be like Palestine next
Full of dead Brown People?
Not really.
You should stick to your area of expertise Malc.!
What drove you off the edge BJ, you were normal on here once.
Wanting to stop Kids being slaughtered is normal 76% of people agree with me
Wanting the slaughter to continue is being "driven off the edge" and "Normal" No
Listen to yourself mate
Yes, the polling on ceasefire is quite strongly in favour, even amongst Tory voters.
Not that either Israel or Hamas are likely to listen to anyone in this country.
"Ceasefire" means both sides stop hostilities.
That means: - All hostages returned unharmed (clearly already impossible, so there's an immediate problem) - Hamas/Islamic Jihad/etc stop firing rockets at Israeli cities - Hamas leadership surrender and are deported to the Hague to face justice for the crimes against humanity they've committed
Once that's done, Israel will have no reason to continue the war anyway.
Of course the majority of people want a ceasefire. The issue is that the above isn't happening anytime soon, or ever.
A ceasefire means both sides cease firing. That's what most people want.
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts) then to say python and still so so.
It just takes one of these machines to have the ability to self improve - then KABOOM
It doesn't even "understand" basic maths yet, let alone know how to improve upon that. I had it the other day arguing black is white over the incorrect details of a tensor, which is trivial maths.
This is the fundamental issue with the idea of LLM being predicted on probabilistic selection of next token.
Arguably it doesn't "understand" anything because it's a goldfish in a bowl with a 5 second memory that happens to be born with an encyclopedia inside its head.
Give it long term memory, and it starts to look shockingly like how human minds work.
An LLM with long term memory accessible the way humans can remember something from 20 years ago at a pinch, starts to look shockingly like us. Give it the ability to improve on its own code and it starts to look very much like us. The human mind is nothing special.
We may be 5-10 years away from this, or some have speculated, Google or OpenAI may be there already but holding back on releasing it for fear of what it will do. But it is coming.
I think need to be careful what I mean by understanding. No its doesn't understand anything in the human sense, but I would argue it has learned very decent set of rules for natural language and that the probabilistic approach then allows for this nice ability to produce psuedo-creative outputs.
My experience working with it, this training approach doesn't seem to have worked very well for ascertaining the rules of maths or coding of maths. There is a reason they hooked it to mathematica, because without, it is actually really bad. Can this be improved, sure, but is the probabilistic LLM approach to trying to model hard science like maths the right approach, I am not convinced. Could there by something similar or better, sure. But we aren't there yet. The human created Mathematica is still vastly superior, and everybody in STEM isn't saying well that all the jobs gone because of Wolfram Alpha / Mathematica (which has been about for what 35 years), rather its a really useful tool.
Yes, I agree with you entirely there. About 50% of my queries to GPT4 involve the Wolfram plugin - e.g. here are a bunch of figures, can you go away and calculate NPV on this project? With Wolfram it gets it right, without, it's a mess. So it has no intelligence or understanding in the human sense, yet.
The interesting question I keep coming back to is, what does intelligence, or as you call it, understanding, really look like, and how will we distinguish it from a really advanced inference model when we reach it?
Most of the people I know working in the AI space at the moment are of the opinion that humans are just really advanced LLMs with gigantic memories and processing power. I suspect this theory will be proven in the next few years - though it's probably harder to disprove.
I like to think back to Data from TNG. Is he really alive, or just a very very fast processor chip with memory access and a 24th century LLM? What makes human consciousness special or unique?
Yes, of course
The obvious inference here is not that AIs are mere auto complete machines which look weirdly human but never can be, it’s that humans are themselves autocomplete machines, and we therefore resist something that looks so spookily similar, and which tells us this unhappy truth. Hence Uncanny Valley
It’s obviously true. We are a bunch of reflexes which react to stimuli, we autocomplete every sentence and every situation
Time for a new party that represents Muslims adequately. We will not win overall but could easily win parliamentary seats. We are currently in active discussions to get this done. Muslim Labour MPs have let the community down. Change is needed.
they will want England to be like Palestine next
Full of dead Brown People?
Not really.
You should stick to your area of expertise Malc.!
What drove you off the edge BJ, you were normal on here once.
Wanting to stop Kids being slaughtered is normal 76% of people agree with me
Wanting the slaughter to continue is being "driven off the edge" and "Normal" No
Listen to yourself mate
Yes, the polling on ceasefire is quite strongly in favour, even amongst Tory voters.
Not that either Israel or Hamas are likely to listen to anyone in this country.
I'm not sure even as simple a concept as ceasefire has a universally accepted interpretation in this instance, with some expecting it to involve return of hostages and others not.
I imagine most people simply interpret it as an operational pause to allow humanitarian aid in.
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Is it like Cleopatra where they'll go 'Oh, we don't really know so they might have been'(which at best is still wishful thinking rather than being able to claim as fact) or is there an actual reason behind the idea? I know at least one Emperor was born in Africa and one was named 'the arab', so did they just go off that?
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts) then to say python and still so so.
It just takes one of these machines to have the ability to self improve - then KABOOM
It doesn't even "understand" basic maths yet, let alone know how to improve upon that. I had it the other day arguing black is white over the incorrect details of a tensor, which is trivial maths.
This is the fundamental issue with the idea of LLM being predicted on probabilistic selection of next token.
Arguably it doesn't "understand" anything because it's a goldfish in a bowl with a 5 second memory that happens to be born with an encyclopedia inside its head.
Give it long term memory, and it starts to look shockingly like how human minds work.
An LLM with long term memory accessible the way humans can remember something from 20 years ago at a pinch, starts to look shockingly like us. Give it the ability to improve on its own code and it starts to look very much like us. The human mind is nothing special.
We may be 5-10 years away from this, or some have speculated, Google or OpenAI may be there already but holding back on releasing it for fear of what it will do. But it is coming.
I think need to be careful what I mean by understanding. No its doesn't understand anything in the human sense, but I would argue it has learned very decent set of rules for natural language and that the probabilistic approach then allows for this nice ability to produce psuedo-creative outputs.
My experience working with it, this training approach doesn't seem to have worked very well for ascertaining the rules of maths or coding of maths. There is a reason they hooked it to mathematica, because without, it is actually really bad. Can this be improved, sure, but is the probabilistic LLM approach to trying to model hard science like maths the right approach, I am not convinced. Could there by something similar or better, sure. But we aren't there yet. The human created Mathematica is still vastly superior, and everybody in STEM isn't saying well that all the jobs gone because of Wolfram Alpha / Mathematica (which has been about for what 35 years), rather its a really useful tool.
Yes, I agree with you entirely there. About 50% of my queries to GPT4 involve the Wolfram plugin - e.g. here are a bunch of figures, can you go away and calculate NPV on this project? With Wolfram it gets it right, without, it's a mess. So it has no intelligence or understanding in the human sense, yet.
The interesting question I keep coming back to is, what does intelligence, or as you call it, understanding, really look like, and how will we distinguish it from a really advanced inference model when we reach it?
Most of the people I know working in the AI space at the moment are of the opinion that humans are just really advanced LLMs with gigantic memories and processing power. I suspect this theory will be proven in the next few years - though it's probably harder to disprove.
I like to think back to Data from TNG. Is he really alive, or just a very very fast processor chip with memory access and a 24th century LLM? What makes human consciousness special or unique?
Yes, of course
The obvious inference here is not that AIs are mere auto complete machines which look weirdly human but never can be, it’s that humans are themselves autocomplete machines, and we therefore resist something that looks so spookily similar, and which tells us this unhappy truth. Hence Uncanny Valley
It’s obviously true. We are a bunch of reflexes which react to stimuli, we autocomplete every sentence and every situation
To me, the philosophical argument is rather mute. The vast majority of the population never even maximise their full intellectual capacity as it is, and huge numbers of jobs are mudane and repetitive, added with poor productivity....the question is really what do countries do when computer systems can replace a large proportion of those workers in the way machines replaced lots of jobs that involved metal bashing etc. And left with the top few % who are at the very high end of the education / intelligence spectrum being the ones making the mega bucks as the ones making these systems or are able to pivot to roles that working on other areas where machines are suboptimal at e.g. fine skilled robotics is still actually very poor, robot hands are actually really rubbish.
I like to think back to Data from TNG. Is he really alive, or just a very very fast processor chip with memory access and a 24th century LLM? What makes human consciousness special or unique?
"That is true. However, I believe my growth as an artificial life-form has reached an impasse. For 34 years, I have endeavored to become more human, to grow beyond my original programming. Still, I am unable to grasp such a basic concept as humour. This emotion chip may be the only answer."
Just a few minutes ago, Imran Shafi was asked about the Eat Out To Help Out scheme put in place by Rishi Sunak when he was chancellor to encourage people back into pubs and restaurants in the summer of 2020.
Remarkably he was asked if chief medical officer Sir Chris Whitty had described it as "Eat Out To Help Out The Virus".
But to me this is quite a confused list, and in some places facing both ways. I'll pick up one or two aspects.
Do we reduce, or increase? The UK has been relatively successful in reducing emissions ("mitigating key risks") - the only sectors not to have made substantial progress are perhaps Transport and Older Housing Stock (especially Owner Occupied), but Rishi Sunak just burnt down the long-term programmes to improve both in order to save his miserable backside in the short-term. And rapid travellers from eg Newcastle to London are now stuck in their aeroplanes, not trains.
Do you want to have more National Parks to preserve the countryside, or concrete the countryside over with more reservoirs and water canals / pipelines?
The numbers are that 12% of the country is already in National Parks, 17-18% is already in Areas of Outstanding Natural Beauty, and that leaves out Green Belt, National Nature Reserves, Local Nature Reserves, Special Protection Areas, Special Areas of Conservation, SSIs etc. And that's all ignoring that eg 1500 sq miles owned by the National Trust / RSPB. And so on.
Plus we have scope to reduce water use by 25% or more just to catch up with European best practice. I'd say that that is the more important goal.
I don't really understand the desire to cleave conservation from farming; I'd suggest they work better together in most places - we can learn for example from the tree belts we see right across Ukraine. Plus if we are about being self-supporting we need to address food security.
Much of the stuff is simple and obvious. Is there anyone in the country who has not switched completely to LCD lightbulbs? I did it in 2013, and it had paid for itself by 2014 in lower electricity bills. Normal loft insulation has been available for free forever under ECO (but Rishi may have burnt that down as well). And so on. I don't see a need for £300bn to be spent nationally on this; it just needs some regulatory changes.
The final question is where is the £710bn to come from? Recall that Social Care, the shredding of Local Government funding, public transport, near-abolition of legal aid, Defence, and a long list of others still require to be addressed. And National Debt here is still high by some international comparisons (eg Germany).
Proportionality means you don’t wipe out several hundred civilians in a building to get to a few Hamas fighters.
Aswell as that are we to believe that Israel destroying half of the homes in Gaza was justified under the Hamas target slogan .
This is a deliberate attempt to force Palestinians to head south and annex the north which is now close to un-inhabitable . No one can look at the devastation and think they were all Hamas targets , it looks more like just carpet bombing .
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Appears as if some of these "right-on" types people think Africa = black people, seems a bit racist to me.
Their obsession with race is very unhealthy IMO.
It's an attempt to create an origin story to legitimate modern Britain, but is incompatible with the alternative Windrush generation narrative, and the reality is that most of the demographic change is even more recent than that.
I like to think back to Data from TNG. Is he really alive, or just a very very fast processor chip with memory access and a 24th century LLM? What makes human consciousness special or unique?
"That is true. However, I believe my growth as an artificial life-form has reached an impasse. For 34 years, I have endeavored to become more human, to grow beyond my original programming. Still, I am unable to grasp such a basic concept as humour. This emotion chip may be the only answer."
I know plenty of humans who haven't been able to grasp basic humour in 34 years, so I woudn't be counting AI out of the race just yet.
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Is it like Cleopatra where they'll go 'Oh, we don't really know so they might have been'(which at best is still wishful thinking rather than being able to claim as fact) or is there an actual reason behind the idea? I know at least one Emperor was born in Africa and one was named 'the arab', so did they just go off that?
One quote from that article
“The ancient biographical collection Historia Augustus explains that Severus was disturbed by the sight of a black person on one occasion, taking his “ominous colour” as a bad omen while on campaign.”
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Just a few minutes ago, Imran Shafi was asked about the Eat Out To Help Out scheme put in place by Rishi Sunak when he was chancellor to encourage people back into pubs and restaurants in the summer of 2020.
Remarkably he was asked if chief medical officer Sir Chris Whitty had described it as "Eat Out To Help Out The Virus".
Shafi confirmed that was the case.
Being locked up in your home with no human contact for months on end was the virus, or at least the thing that made me seriously ill.
Just a few minutes ago, Imran Shafi was asked about the Eat Out To Help Out scheme put in place by Rishi Sunak when he was chancellor to encourage people back into pubs and restaurants in the summer of 2020.
Remarkably he was asked if chief medical officer Sir Chris Whitty had described it as "Eat Out To Help Out The Virus".
Shafi confirmed that was the case.
I always thought of it as "Eat Out to Spread COVID About".
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Appears as if some of these "right-on" types people think Africa = black people, seems a bit racist to me.
Their obsession with race is very unhealthy IMO.
It all seems a bit silly really. How does it help the Black person on the Clapham omnibus if some ancient Roman emperor no-one has heard of turns out to be Black? Or harm them if he wasn't?
I am surprised Leon hasn't commented on the new exhibition that replaced the Medicine Man at the Wellcome Collection...it literally includes a bearded lady.
Featuring over 200 items, including historical objects, artworks, films and new commissions, we consider the influence of morality, status, health, age, race and gender on the evolution of ideas about beauty. We invite you to question established norms and reflect on more inclusive definitions of beauty.
This stuff was once "out there" art, now it just seems so passe.
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts, which it gets wrong and I could write it myself very quickly) then to say python (which GPT3 could do) and result of this is still so so.
Even if you give it the whole page with all the details of what the various symbols mean, unless its trivial, its not great at it. And if its trivial, a human coder can do it just as fast anyway.
Again, will this get better and better, absolutely.
Tbh I know fuck all about coding, and care less (other than it’s huge secondary impact on our lives, of course)
What I do know and care about is the arts, writing, design, advertising, drawing, thinking, all the interesting stuff the mind does - and if GPT5 is
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts) then to say python and still so so.
It just takes one of these machines to have the ability to self improve - then KABOOM
It doesn't even "understand" basic maths yet, let alone know how to improve upon that. I had it the other day arguing black is white over the incorrect details of a tensor, which is trivial maths.
This is the fundamental issue with the idea of LLM being predicted on probabilistic selection of next token.
Does a calculator “understand” maths? Not in the sense you mean, probably not. Surely not
How do you know?
It took Douglas Adams to demonstrate that the world is run by mice.
Just a few minutes ago, Imran Shafi was asked about the Eat Out To Help Out scheme put in place by Rishi Sunak when he was chancellor to encourage people back into pubs and restaurants in the summer of 2020.
Remarkably he was asked if chief medical officer Sir Chris Whitty had described it as "Eat Out To Help Out The Virus".
Shafi confirmed that was the case.
I always thought of it as "Eat Out to Spread COVID About".
Boris has been an idiot when it comes to many things, eg party gate, Carrie, etc, but in this he was totally right. Would we rather not have a PM asking these pointed, direct questions, when we are about to lockdown the entire country, and saddle us all with a trillion quid of debt?
So he said it bluntly. Even better. Make the point
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts) then to say python and still so so.
It just takes one of these machines to have the ability to self improve - then KABOOM
It doesn't even "understand" basic maths yet, let alone know how to improve upon that. I had it the other day arguing black is white over the incorrect details of a tensor, which is trivial maths.
This is the fundamental issue with the idea of LLM being predicted on probabilistic selection of next token.
Arguably it doesn't "understand" anything because it's a goldfish in a bowl with a 5 second memory that happens to be born with an encyclopedia inside its head.
Give it long term memory, and it starts to look shockingly like how human minds work.
An LLM with long term memory accessible the way humans can remember something from 20 years ago at a pinch, starts to look shockingly like us. Give it the ability to improve on its own code and it starts to look very much like us. The human mind is nothing special.
We may be 5-10 years away from this, or some have speculated, Google or OpenAI may be there already but holding back on releasing it for fear of what it will do. But it is coming.
I think need to be careful what I mean by understanding. No its doesn't understand anything in the human sense, but I would argue it has learned very decent set of rules for natural language and that the probabilistic approach then allows for this nice ability to produce psuedo-creative outputs.
My experience working with it, this training approach doesn't seem to have worked very well for ascertaining the rules of maths or coding of maths. There is a reason they hooked it to mathematica, because without, it is actually really bad. Can this be improved, sure, but is the probabilistic LLM approach to trying to model hard science like maths the right approach, I am not convinced. Could there by something similar or better, sure. But we aren't there yet. The human created Mathematica is still vastly superior, and everybody in STEM isn't saying well that all the jobs gone because of Wolfram Alpha / Mathematica (which has been about for what 35 years), rather its a really useful tool.
Yes, I agree with you entirely there. About 50% of my queries to GPT4 involve the Wolfram plugin - e.g. here are a bunch of figures, can you go away and calculate NPV on this project? With Wolfram it gets it right, without, it's a mess. So it has no intelligence or understanding in the human sense, yet.
The interesting question I keep coming back to is, what does intelligence, or as you call it, understanding, really look like, and how will we distinguish it from a really advanced inference model when we reach it?
Most of the people I know working in the AI space at the moment are of the opinion that humans are just really advanced LLMs with gigantic memories and processing power. I suspect this theory will be proven in the next few years - though it's probably harder to disprove.
I like to think back to Data from TNG. Is he really alive, or just a very very fast processor chip with memory access and a 24th century LLM? What makes human consciousness special or unique?
Yes, of course
The obvious inference here is not that AIs are mere auto complete machines which look weirdly human but never can be, it’s that humans are themselves autocomplete machines, and we therefore resist something that looks so spookily similar, and which tells us this unhappy truth. Hence Uncanny Valley
It’s obviously true. We are a bunch of reflexes which react to stimuli, we autocomplete every sentence and every situation
To me, the philosophical argument is rather mute.
Most people are content to misumderstand what 'moot' means, without substituting the wrong word as well.
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
Used Apple Maps on my iPhone for the first time in years this weekend (I always used Google Maps before).
It is stunning. Far better interface than Google Maps (which has essentially stood still for five years).
Clearer, easier to use, and the integration with the Apple Watch is absolutely superb.
Apple just need to produce a search engine, then it will be possible to remove Google from my life once and for all – it's a dull, boring, ugly interface that has barely progressed.
That said, blunt as it was put it's not as though Boris took a laissez-faire approach to Covid regulations, the UK was not wide open or anything. Indeed, if earlier reports were correct Rishi was more of a 'reduce restrictions' guy than Boris was, once things had been in place for awhile.
Just a few minutes ago, Imran Shafi was asked about the Eat Out To Help Out scheme put in place by Rishi Sunak when he was chancellor to encourage people back into pubs and restaurants in the summer of 2020.
Remarkably he was asked if chief medical officer Sir Chris Whitty had described it as "Eat Out To Help Out The Virus".
Shafi confirmed that was the case.
Being locked up in your home with no human contact for months on end was the virus, or at least the thing that made me seriously ill.
And the same went for many others. Sunak is and was shit, but Eoto was one of the few good ideas he had – helped an industry that was scythed to its knees due to government fiat, and was in place in the middle of summer when covid infections were at record lows for the pandemic.
I think AlanBrookes approach is not dissimilar to Biden's/China tbh - positive investment into various projects rather than thou shalt not have a gas boiler post 2035, ban on ICE vehicles favoured by the EU and accelerated till Rishis intervention by the UK.
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Appears as if some of these "right-on" types people think Africa = black people, seems a bit racist to me.
Their obsession with race is very unhealthy IMO.
It all seems a bit silly really. How does it help the Black person on the Clapham omnibus if some ancient Roman emperor no-one has heard of turns out to be Black? Or harm them if he wasn't?
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
Used Apple Maps on my iPhone for the first time in years this weekend (I always used Google Maps before).
It is stunning. Far better interface than Google Maps (which has essentially stood still for five years).
Clearer, easier to use, and the integration with the Apple Watch is absolutely superb.
Apple just need to produce a search engine, then it will be possible to remove Google from my life once and for all – it's a dull, boring, ugly interface that has barely progressed.
Yep
Apple Maps has quietly overtaken Google Maps in many ways
Google Maps has always been my go-to, for a decade. Now I often use Apple
Google is having its own Facebook pants-down moment
And yet BOTH Apple and google have been blindsided on AI by Microsoft and OpenAI, and this despite Google buying Deepmind one of the best AI set-ups in the biz
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Appears as if some of these "right-on" types people think Africa = black people, seems a bit racist to me.
Their obsession with race is very unhealthy IMO.
It all seems a bit silly really. How does it help the Black person on the Clapham omnibus if some ancient Roman emperor no-one has heard of turns out to be Black? Or harm them if he wasn't?
I believe those who push this, it often comes from white guilt about historic (racist) portrayal of black Africans being uneducated, unsophisticated savages, so they are always trying to find historic examples to the contrary in order correct the narrative. The problem is just making shit up doesn't help anybody.
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
Used Apple Maps on my iPhone for the first time in years this weekend (I always used Google Maps before).
It is stunning. Far better interface than Google Maps (which has essentially stood still for five years).
Clearer, easier to use, and the integration with the Apple Watch is absolutely superb.
Apple just need to produce a search engine, then it will be possible to remove Google from my life once and for all – it's a dull, boring, ugly interface that has barely progressed.
How it happened
“A decade after a disastrous launch, is Apple Maps finally good?”
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
Used Apple Maps on my iPhone for the first time in years this weekend (I always used Google Maps before).
It is stunning. Far better interface than Google Maps (which has essentially stood still for five years).
Clearer, easier to use, and the integration with the Apple Watch is absolutely superb.
Apple just need to produce a search engine, then it will be possible to remove Google from my life once and for all – it's a dull, boring, ugly interface that has barely progressed.
Apple currently has a $20 billion+ year incentive from Google not to develop a better search engine.
We were discussing the impact of AI on the global south yesterday. I’ve just read a cogent article saying 5m coding jobs in India will disappear immediately - ie in the next two years. Not in a decade. From now on
I use ChatGPT and CoPilot every day to assist with coding. Now I am not doing run of mill boilerplate stuff, but it still makes absolutely loads of mistakes. I would say at the moment, it is more like an a combination of decent auto-complete (like you do with text messages) and an advanced search for stack overflow (which all coders have used for donkeys years to get an idealised solution to common problems). Once you get too far from tasks that are easily found there, it starts to break down.
In summary, will it cause lots of job losses a few years down the line, yes. I wouldn't want to be a random coder starting out their career now. Would I trust it to go from 10 coders to 1-3 coders doing the same job, not yet, not unless you want massively buggy code.
Indeed. I use ChatGPT and AWS's CodeWhisperer, and they're great tools, but that's all they are - tools. You have to be really careful how you use them though, and you still have to understand what you are doing. I've already been caught out once or twice by suggested code that initially looked perfectly good but turned out to contain some stupid bug that I missed on checking through it.
Lots of these prognoses are predicated on another leap by AI in the next year. GPT5 etc
And if the rumours are right, and GPT5 is to GPT4 what GPT4 was to 3, then they are right to be “alarmist”
And that’s just OpenAI
GPT3 -> GPT4 for coding wasn't that big of advance. It was evolution, rather than revolution. You could say well now you can show it a picture of some maths and it will code that function, thats new, but again, its very very hit and miss. I have been trying this to speed up implementing academic papers and its really not that great. I find I have to convert image to latex (a kinda of programming language to create scientific texts) then to say python and still so so.
It just takes one of these machines to have the ability to self improve - then KABOOM
It doesn't even "understand" basic maths yet, let alone know how to improve upon that. I had it the other day arguing black is white over the incorrect details of a tensor, which is trivial maths.
This is the fundamental issue with the idea of LLM being predicted on probabilistic selection of next token.
Arguably it doesn't "understand" anything because it's a goldfish in a bowl with a 5 second memory that happens to be born with an encyclopedia inside its head.
Give it long term memory, and it starts to look shockingly like how human minds work.
An LLM with long term memory accessible the way humans can remember something from 20 years ago at a pinch, starts to look shockingly like us. Give it the ability to improve on its own code and it starts to look very much like us. The human mind is nothing special.
We may be 5-10 years away from this, or some have speculated, Google or OpenAI may be there already but holding back on releasing it for fear of what it will do. But it is coming.
I think need to be careful what I mean by understanding. No its doesn't understand anything in the human sense, but I would argue it has learned very decent set of rules for natural language and that the probabilistic approach then allows for this nice ability to produce psuedo-creative outputs.
My experience working with it, this training approach doesn't seem to have worked very well for ascertaining the rules of maths or coding of maths. There is a reason they hooked it to mathematica, because without, it is actually really bad. Can this be improved, sure, but is the probabilistic LLM approach to trying to model hard science like maths the right approach, I am not convinced. Could there by something similar or better, sure. But we aren't there yet. The human created Mathematica is still vastly superior, and everybody in STEM isn't saying well that all the jobs gone because of Wolfram Alpha / Mathematica (which has been about for what 35 years), rather its a really useful tool.
Yes, I agree with you entirely there. About 50% of my queries to GPT4 involve the Wolfram plugin - e.g. here are a bunch of figures, can you go away and calculate NPV on this project? With Wolfram it gets it right, without, it's a mess. So it has no intelligence or understanding in the human sense, yet.
The interesting question I keep coming back to is, what does intelligence, or as you call it, understanding, really look like, and how will we distinguish it from a really advanced inference model when we reach it?
Most of the people I know working in the AI space at the moment are of the opinion that humans are just really advanced LLMs with gigantic memories and processing power. I suspect this theory will be proven in the next few years - though it's probably harder to disprove.
I like to think back to Data from TNG. Is he really alive, or just a very very fast processor chip with memory access and a 24th century LLM? What makes human consciousness special or unique?
Yes, of course
The obvious inference here is not that AIs are mere auto complete machines which look weirdly human but never can be, it’s that humans are themselves autocomplete machines, and we therefore resist something that looks so spookily similar, and which tells us this unhappy truth. Hence Uncanny Valley
It’s obviously true. We are a bunch of reflexes which react to stimuli, we autocomplete every sentence and every situation
To me, the philosophical argument is rather mute.
Most people are content to misumderstand what 'moot' means, without substituting the wrong word as well.
I apologise for the mild form of dyslexia that I suffer from. If you see my posts, I regularly do this, which I often try and go back and edit when I notice it.
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Is it like Cleopatra where they'll go 'Oh, we don't really know so they might have been'(which at best is still wishful thinking rather than being able to claim as fact) or is there an actual reason behind the idea? I know at least one Emperor was born in Africa and one was named 'the arab', so did they just go off that?
One quote from that article
“The ancient biographical collection Historia Augustus explains that Severus was disturbed by the sight of a black person on one occasion, taking his “ominous colour” as a bad omen while on campaign.”
I mean, I know, but. Lol
Ouch.
I mean, no one expects everyone to be an expert to know that level of detail, but does the story provide an explanation from the BBC and museums as to how they made the error? Because if it is as simple as 'Wikipedia says he was born in Africa and we just assumed' that is bad, but if that is not the reason they thought it then what was the reason?
Boris has been an idiot when it comes to many things, eg party gate, Carrie, etc, but in this he was totally right. Would we rather not have a PM asking these pointed, direct questions, when we are about to lockdown the entire country, and saddle us all with a trillion quid of debt?
So he said it bluntly. Even better. Make the point
In any case the Tories had a much better Truss-shaped weapon lined up to destroy the economy.
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
Used Apple Maps on my iPhone for the first time in years this weekend (I always used Google Maps before).
It is stunning. Far better interface than Google Maps (which has essentially stood still for five years).
Clearer, easier to use, and the integration with the Apple Watch is absolutely superb.
Apple just need to produce a search engine, then it will be possible to remove Google from my life once and for all – it's a dull, boring, ugly interface that has barely progressed.
Google is a mess. Its incredibly inefficient. They have become so large, the left hand doesn't know what the right hand is doing. Obviously they have become rather infamous for making products which they quickly ditch, but lesser know are the number of projects they end up doing which are direct copies of ones they already have in the works because the communication and leadership are not keeping on top of things.
They of course produced the break through paper which LLMs are based upon, the Attention Based Transformer, in 2017, but decided wasn't really that important, while Google Brain and DeepMind dropped the ball (rather than now, working together). By their own admission, they could have produced ChatGPT by 2018/19 easily if they had really focused on it. They had the compute power, the engineering might etc. OpenAI had to get daddy Microsoft to do a special deal with NVidia to get them the kit to train up their models.
Storm Ciaran evening update, and the headline is that things have got worse.
Top gusts along the coast in high-res Arpege model now nudging 100mph. Inland up to 90mph in the SW, 80mph in the South.
GFS less intense on the coast but gusts up to 70mph getting inland as far as London.
UKMO pretty strong along the coast but not getting much inland.
Some ensemble members going full 1987 with 100mph+ in central Kent.
Some pretty serious rainfall amounts on the way too. And trees still have their leaves.
In 1987, recovery after the storm was faster than expected because the CEGB (Central Electricity Generating Board) bussed staff and equipment down from up north to repair power lines and get the grid back up and running. Let's hope there is a similar contingency plan in someone's desk drawer.
A Roman emperor has been listed as “black Briton” by the BBC and some UK museums, despite not being black.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
Appears as if some of these "right-on" types people think Africa = black people, seems a bit racist to me.
Their obsession with race is very unhealthy IMO.
It all seems a bit silly really. How does it help the Black person on the Clapham omnibus if some ancient Roman emperor no-one has heard of turns out to be Black? Or harm them if he wasn't?
I guess it's meant to be a way of showing how things used to be better (look, even a Roman Emperor who was in Britain was black), but are now worse, despite the fact a lot of progress has been made just in recent decades alone (look at ethnic minority representation in parliament for example)?
"The cost of processing a card payment has increased by 600 per cent in the past nine years, according to campaign group Axe the Card Tax."
Doubt.
They says costs are 2.5% now so back in 2014 costs were 0.004%?
I call fucking bullshit.
0.4%, Shirley? 600 percent is six times.
I still call fucking bullshit.
Card costs have come down.
Numerical calculation out by a factor of 100 but fronting it out and doubling down on your original claim, bold. Have you considered a career in politics, young man?
Have we all got our credit cards ready for the big Apple event in a few hours....I am sure TSE will be down £4-5k come tomorrow.
I recently spent over 2k on the latest MacBook Air.
I am all shopped out.
Famous last words.
Well that was a bit stupid ;-) ....I am sure after Tim Apple has got your all hot and bothered with some new sexy M3 chipped laptop, TSE jnr will be getting the MacBook Air from Santa....
Used Apple Maps on my iPhone for the first time in years this weekend (I always used Google Maps before).
It is stunning. Far better interface than Google Maps (which has essentially stood still for five years).
Clearer, easier to use, and the integration with the Apple Watch is absolutely superb.
Apple just need to produce a search engine, then it will be possible to remove Google from my life once and for all – it's a dull, boring, ugly interface that has barely progressed.
Yep
Apple Maps has quietly overtaken Google Maps in many ways
Google Maps has always been my go-to, for a decade. Now I often use Apple
Google is having its own Facebook pants-down moment
And yet BOTH Apple and google have been blindsided on AI by Microsoft and OpenAI, and this despite Google buying Deepmind one of the best AI set-ups in the biz
Thinking of deleting Google Maps. It is drab, and has some infuriating flaws. One that comes to mind is that it doesn't display the opening times of pubs if they also have rooms (it has a dialogue box to reserve a room instead) – so is absolutely useless when you are hiking/cycling and just want to find a pub that is open and serving beer and food at 3.30pm in the Chilterns on a Tuesday etc.
"The cost of processing a card payment has increased by 600 per cent in the past nine years, according to campaign group Axe the Card Tax."
Doubt.
They says costs are 2.5% now so back in 2014 costs were 0.004%?
I call fucking bullshit.
0.4%, Shirley? 600 percent is six times.
I still call fucking bullshit.
Card costs have come down.
Numerical calculation out by a factor of 100 but fronting it out and doubling down on your original claim, bold. Have you considered a career in politics, young man?
It was a lack of self confidence that stopped me going into politics.
Comments
"Paul Bristow: Ministerial aide sacked after call for Israel-Gaza ceasefire"
https://space.stackexchange.com/questions/45617/why-is-it-easier-to-escape-the-solar-system-than-get-to-mercury-or-the-sun
@joannaccherry
Those posting snide comments about Ash Regan defection to Alba should stop & reflect on whether they have played a part in splitting the independence movement. The world & our country face such huge problems I wd like to see reconciliation not more division
2:30 PM · Oct 28, 2023"
https://twitter.com/joannaccherry/status/1718258834462884288
My experience working with it, this training approach doesn't seem to have worked very well for ascertaining the rules of maths or coding of maths. There is a reason they hooked it to mathematica, because without, it is actually really bad. Can this be improved, sure, but is the probabilistic LLM approach to trying to model hard science like maths the right approach, I am not convinced. Could there by something similar or better, sure. But we aren't there yet. The human created Mathematica is still vastly superior, and everybody in STEM isn't saying well that all the jobs gone because of Wolfram Alpha / Mathematica (which has been about for what 35 years), rather its a really useful tool.
Wanting the slaughter to continue is being "driven off the edge" and "Normal" No
Listen to yourself mate
Not that either Israel or Hamas are likely to listen to anyone in this country.
I think Starmer could make a virtue of this, particularly after the Bristow sacking. It's in some ways a matter of conscience because the debate isn't really over British policies per se. Let people voice their own views on this so long as they avoid Antisemitism or Islamophobia.
Furthermore, the argument presented in the paper is that isn't just too many mickey mouse degrees, but again the US has plenty of those as well.
Now, I might say that its rather circular argument, you have lots of STEM graduates, you more likely to get high tech industry, which then spills over across the board. That is certainly what has happened in the US cities like San Francisco.
I think we all know for too long productivity has also been a massive issue in the UK.
That means:
- All hostages returned unharmed (clearly already impossible, so there's an immediate problem)
- Hamas/Islamic Jihad/etc stop firing rockets at Israeli cities
- Hamas leadership surrender and are deported to the Hague to face justice for the crimes against humanity they've committed
Once that's done, Israel will have no reason to continue the war anyway.
Of course the majority of people want a ceasefire. The issue is that the above isn't happening anytime soon, or ever.
“The value of your Education and the income it may generate can go down as well as up and is not guaranteed at any time”.
https://www.thetimes.co.uk/article/could-viagra-help-to-prevent-dementia-j7z830k67
I am all shopped out.
Famous last words.
"Black Met officers urged by colleagues to donate to fund for sacked PCs
Exclusive: Black officers complain of collection for men who lied about the stop and search of black athletes"
Isn't this the sort of thing The Home Secretary should be criticising? Or is there some reason that she might not want to, that I can't think of right now...
Labour leads by 20% nationally.
Westminster VI (29 October):
Labour 45% (+1)
Conservative 25% (-1)
Liberal Democrat 13% (–)
Reform UK 7% (-1)
Green 6% (+2)
Scottish National Party 3% (+1)
https://twitter.com/RedfieldWilton/status/1719036442494586948
The interesting question I keep coming back to is, what does intelligence, or as you call it, understanding, really look like, and how will we distinguish it from a really advanced inference model when we reach it?
Most of the people I know working in the AI space at the moment are of the opinion that humans are just really advanced LLMs with gigantic memories and processing power. I suspect this theory will be proven in the next few years - though it's probably harder to disprove.
I like to think back to Data from TNG. Is he really alive, or just a very very fast processor chip with memory access and a 24th century LLM? What makes human consciousness special or unique?
I recently bought him an Alienware gaming laptop.
LD have been very stable in recent polls across most pollsters. Feels like the Dutch salute has been done, the top remains up, people have made a settled choice. I'd expected a fall after the Mid-Beds result but nothing obvious.
"More than 17% of people in Mr Bristow's Peterborough constituency are Muslim, according to the latest census.
It is a marginal seat, which he won from Labour with a majority of 2,580 at the last general election."
They know that this is a much bigger issue for Labour and Starmer sacking ministers will lead to much more trouble for the party.
Lucius Septimius Severus died on campaign in Britain in 211AD , and has been included in teaching material and children’s books alongside influential black Britons such as Mary Seacole and Olaudah Equiano.
https://www.telegraph.co.uk/news/2023/10/30/bbc-hails-roman-emperor-black-briton/
Appears as if some of these "right-on" types people think Africa = black people, seems a bit racist to me.
Same as with all but one administration in my lifetime.
During the Cheney-Bush administration the head honcho graciously gave quasi-equal billing to HIS sidekick, W.
The obvious inference here is not that AIs are mere auto complete machines which look weirdly human but never can be, it’s that humans are themselves autocomplete machines, and we therefore resist something that looks so spookily similar, and which tells us this unhappy truth. Hence Uncanny Valley
It’s obviously true. We are a bunch of reflexes which react to stimuli, we autocomplete every sentence and every situation
“We are killing the patient to tackle the tumour”
Next line: Why are we destroying the economy for people who are going to die anyway?
https://twitter.com/nickeardleybbc/status/1719028306907955345
Remarkably he was asked if chief medical officer Sir Chris Whitty had described it as "Eat Out To Help Out The Virus".
Shafi confirmed that was the case.
But to me this is quite a confused list, and in some places facing both ways. I'll pick up one or two aspects.
Do we reduce, or increase? The UK has been relatively successful in reducing emissions ("mitigating key risks") - the only sectors not to have made substantial progress are perhaps Transport and Older Housing Stock (especially Owner Occupied), but Rishi Sunak just burnt down the long-term programmes to improve both in order to save his miserable backside in the short-term. And rapid travellers from eg Newcastle to London are now stuck in their aeroplanes, not trains.
Do you want to have more National Parks to preserve the countryside, or concrete the countryside over with more reservoirs and water canals / pipelines?
The numbers are that 12% of the country is already in National Parks, 17-18% is already in Areas of Outstanding Natural Beauty, and that leaves out Green Belt, National Nature Reserves, Local Nature Reserves, Special Protection Areas, Special Areas of Conservation, SSIs etc. And that's all ignoring that eg 1500 sq miles owned by the National Trust / RSPB. And so on.
Plus we have scope to reduce water use by 25% or more just to catch up with European best practice. I'd say that that is the more important goal.
I don't really understand the desire to cleave conservation from farming; I'd suggest they work better together in most places - we can learn for example from the tree belts we see right across Ukraine. Plus if we are about being self-supporting we need to address food security.
Much of the stuff is simple and obvious. Is there anyone in the country who has not switched completely to LCD lightbulbs? I did it in 2013, and it had paid for itself by 2014 in lower electricity bills. Normal loft insulation has been available for free forever under ECO (but Rishi may have burnt that down as well). And so on. I don't see a need for £300bn to be spent nationally on this; it just needs some regulatory changes.
The final question is where is the £710bn to come from? Recall that Social Care, the shredding of Local Government funding, public transport, near-abolition of legal aid, Defence, and a long list of others still require to be addressed. And National Debt here is still high by some international comparisons (eg Germany).
Aswell as that are we to believe that Israel destroying half of the homes in Gaza was justified under the Hamas target slogan .
This is a deliberate attempt to force Palestinians to head south and annex the north which is now close to un-inhabitable . No one can look at the devastation and think they were all Hamas targets , it looks more like just carpet bombing .
“The ancient biographical collection Historia Augustus explains that Severus was disturbed by the sight of a black person on one occasion, taking his “ominous colour” as a bad omen while on campaign.”
I mean, I know, but. Lol
https://www.msn.com/en-gb/news/uknews/plaque-honouring-the-first-black-briton-is-removed-after-dna-analysis-finds-she-was-most-likely-from-cyprus/ar-AA1iQxIm
"Plaque honouring the 'first black Briton' is removed after DNA analysis finds she was 'most likely from Cyprus'"
Featuring over 200 items, including historical objects, artworks, films and new commissions, we consider the influence of morality, status, health, age, race and gender on the evolution of ideas about beauty. We invite you to question established norms and reflect on more inclusive definitions of beauty.
This stuff was once "out there" art, now it just seems so passe.
It took Douglas Adams to demonstrate that the world is run by mice.
Top gusts along the coast in high-res Arpege model now nudging 100mph. Inland up to 90mph in the SW, 80mph in the South.
GFS less intense on the coast but gusts up to 70mph getting inland as far as London.
UKMO pretty strong along the coast but not getting much inland.
Some ensemble members going full 1987 with 100mph+ in central Kent.
Some pretty serious rainfall amounts on the way too. And trees still have their leaves.
So he said it bluntly. Even better. Make the point
It is stunning. Far better interface than Google Maps (which has essentially stood still for five years).
Clearer, easier to use, and the integration with the Apple Watch is absolutely superb.
Apple just need to produce a search engine, then it will be possible to remove Google from my life once and for all – it's a dull, boring, ugly interface that has barely progressed.
Fees for contactless are rising massively for small businesses.
https://www.thisismoney.co.uk/money/cardsloans/article-12682825/Pay-cash-spare-crippling-card-fees-plead-traders.html
That said, blunt as it was put it's not as though Boris took a laissez-faire approach to Covid regulations, the UK was not wide open or anything. Indeed, if earlier reports were correct Rishi was more of a 'reduce restrictions' guy than Boris was, once things had been in place for awhile.
https://www.telegraph.co.uk/news/2023/10/30/bbc-hails-roman-emperor-black-briton/
https://www.agecrofthall.org/single-post/jacques-francis-salvage-diver-diving-for-the-king
Apple Maps has quietly overtaken Google Maps in many ways
Google Maps has always been my go-to, for a decade. Now I often use Apple
Google is having its own Facebook pants-down moment
And yet BOTH Apple and google have been blindsided on AI by Microsoft and OpenAI, and this despite Google buying Deepmind one of the best AI set-ups in the biz
“A decade after a disastrous launch, is Apple Maps finally good?”
https://www.theguardian.com/travel/2023/aug/09/apple-maps-cycling-transport-google?CMP=Share_iOSApp_Other
Doubt.
I mean, no one expects everyone to be an expert to know that level of detail, but does the story provide an explanation from the BBC and museums as to how they made the error? Because if it is as simple as 'Wikipedia says he was born in Africa and we just assumed' that is bad, but if that is not the reason they thought it then what was the reason?
They of course produced the break through paper which LLMs are based upon, the Attention Based Transformer, in 2017, but decided wasn't really that important, while Google Brain and DeepMind dropped the ball (rather than now, working together). By their own admission, they could have produced ChatGPT by 2018/19 easily if they had really focused on it. They had the compute power, the engineering might etc. OpenAI had to get daddy Microsoft to do a special deal with NVidia to get them the kit to train up their models.
I call fucking bullshit.
Card costs have come down.
Doesn't appear Bibi is taking any notice of Khan, Burnham, Yousaf, Bristow, etc.
"That is what suspended Labour MP said.. what is wrong with it, Starmer needs a good kick in the bollocks"
I am way too modest for politics.