There's an interesting sentence at the beginning of this Substack article:
"Can the UK ban building superintelligent AI? In a refreshingly frank interview, the Minister for AI Kanishika Naryan observed we effectively already have. Policy on copyright and compute is such that there’s no realistic way to build a frontier AI model in the country that invented computers."
One other point. Musk has stated that in the near(ish) future most compute will reside in space (away from the Regulating Democrats). While he is undoubtedly talking his book, he does have a track record of delivery.
I do think the US and Chinese governments should agree to temporarily pause research into more advanced AI, ideally before March 2027, and agree a framework for regulating frontier AI. I believe we are very close to the point where AI could cause catastrophic risks, either by escaping from human control or by a human hacker or terrorist using AI to unleash chaos, and I think it's astonishing that the AI industry is basically completely unregulated.
Noah Smith (Noahpinion) wrote recently about the possibility of AI-enabled bioterrorism:
It's so rare to see a British politician who realises that AI is something they should be paying attention to. Apart from you, the only MPs I've seen writing about it are Rishi Sunak (Conservative), Danny Kruger (Reform) and Darren Jones (Labour).
EDIT: Thank you for being so on top of things and for writing this. I hope Andy Burnham reads it (and Kemi Badenoch, Nigel Farage, Ed Davey and Zack Polanski). I would give it 100 likes if I could! ♥️💯
Lots of good conversations to be had around AI and the future, thanks for contributing your "conversation starter".
A few more thoughts.
Data. Lots of Data.
- Way back when data storage was scarce and processing power was often even more of a limitation, real focus was placed on data structure design and keeping data clean (eg knowing what the current version is, deleting old and obsolete versions).
- Now, though, for years the cost of storage and of accessing data from storage has been far lower than the cost of having humans actually manage and "keep clean" all that data.
- Recognising that large global businesses tend to have access to new technology first, why is it, then, that few of the largest mature corporations have really shown giant strides with AI?
- One key reason is, yes, data. I know of several that realised several years ago that to fully benefit from AI, they would now have to go through the exercise of cleaning up their data, and when they started to really look at it, they found that they had a LOT of "version control" to do!
- So, the "down and dirty" work of data clean-up has been underway for large corporates for some time, and for mid-size and SMEs some are only just now recognising the drag this will place on their ability to innovate their business with AI.
- In the meantime, we can notice how much "AI middleware" is emerging, promising that it can find data, patterns, and knowledge from within our "data swamps" (as opposed to data lakes). This is very tempting, given it is so difficult to clean up those swamps.
What is the (economic) endgame. Growth or just M&A?
- We are told today that Anthropic may value themselves at $1.5tn for their IPO.
- That is one big number, but as valuation is always (some fundamentals never change) based on market estimates of future profits, what does that mean in terms of revenue and profit growth that will have to be achieved by Anthropic for IPO buyers to ever make any return?
- Well, Noah Smith has published an excellent thought piece on it. In short, Anthropic can't possibly generate real productivity based growth to justify their valuation, they will have to do it by using competitive advantage to gobble up existing companies with their capital. In short, an M&A spree beyond anything we have ever seen.
- Nearly 100 years ago (1930), Keynes forecast that increasing economic productivity would have increased so much (4-8x) within the next 100 years (it has increased 5.5x) that a real problem for society would be how we meaningfully fill our time, so the workweek would be cut to around 15 hours so that we could spread out the work to be done enough so that we could all find meaning in that.
- Despite 5.5x productivity growth, many are working longer and harder than ever before, though the developed world seems not to want to acknowledge that, along with social inequality, we have increasingly large swathes of society who are doing no work at all.
- For those who have the privilege and/or choice to work as much as they want, not as much as they have to, meaning in life can absolutely come (at least in part) from work and that is why some keep doing at least some work (I count myself in this number), but at the same time there are increasingly large numbers who want/need to work but can't find work. This may well increase due to AI, leading not just to a loss of meaning in life, but, more directly, to social unrest (as history shows us happens as inequality grows to extremes).
- Anyone reading this can tell that I do look at a lot of numbers and behaviour through the eyes of an economist, so for those on the far end of the free-market side (eg Jacob Rees-Mogg recently declaring that he does not believe in redistribution of income or wealth at any level at all, people should work for a living, etc.), do we want to live as the wealthier do in the most unequal societies, with social unrest meaning we have to live behind fences and gates with security stopping us from having our bubble popped?
Thanks again, Neil, your conversation starter got me thinking.
I know my views are on the more alarmist side of things, but I honestly believe that within a few months AI may be powerful enough to cause human extinction, and that society should prioritise thinking about that and making sure it doesn't happen, before worrying about economic effects. To me "jobs, inequality etc." are long term issues we can worry about later and preventing a catastrophe that might wipe out civilisation, or humanity, is the short term issue we should worry about right now.
Agreed, eg Dario Amodei's essays etc where even he talks about a 15% chance of this. Sometimes reading science fiction (or watching science fiction movies) is scarily prescient too
There's an interesting sentence at the beginning of this Substack article:
"Can the UK ban building superintelligent AI? In a refreshingly frank interview, the Minister for AI Kanishika Naryan observed we effectively already have. Policy on copyright and compute is such that there’s no realistic way to build a frontier AI model in the country that invented computers."
https://www.samdumitriu.com/p/will-data-centres-drink-us-dry
(Sorry for spamming your comments. Going to go and touch grass now)
One other point. Musk has stated that in the near(ish) future most compute will reside in space (away from the Regulating Democrats). While he is undoubtedly talking his book, he does have a track record of delivery.
You should check out the Threading the Needle Substack by Anton Leicht. It's all about what governments should do in the age of AI.
https://writing.antonleicht.me/archive
I do think the US and Chinese governments should agree to temporarily pause research into more advanced AI, ideally before March 2027, and agree a framework for regulating frontier AI. I believe we are very close to the point where AI could cause catastrophic risks, either by escaping from human control or by a human hacker or terrorist using AI to unleash chaos, and I think it's astonishing that the AI industry is basically completely unregulated.
Noah Smith (Noahpinion) wrote recently about the possibility of AI-enabled bioterrorism:
https://www.noahpinion.blog/p/heres-how-were-all-going-to-die
EDIT: Oops I see you already linked to that post, sorry!
> As someone who grew up on 1970s / 80s books about the space age - which promised us that we would all be living on the moon by now - I say: at last!
Did you ever read The Usborne Book of the Future?
https://www.amazon.co.uk/Book-Future-Kenneth-Gatland/dp/1803709545
It's so rare to see a British politician who realises that AI is something they should be paying attention to. Apart from you, the only MPs I've seen writing about it are Rishi Sunak (Conservative), Danny Kruger (Reform) and Darren Jones (Labour).
EDIT: Thank you for being so on top of things and for writing this. I hope Andy Burnham reads it (and Kemi Badenoch, Nigel Farage, Ed Davey and Zack Polanski). I would give it 100 likes if I could! ♥️💯
Lots of good conversations to be had around AI and the future, thanks for contributing your "conversation starter".
A few more thoughts.
Data. Lots of Data.
- Way back when data storage was scarce and processing power was often even more of a limitation, real focus was placed on data structure design and keeping data clean (eg knowing what the current version is, deleting old and obsolete versions).
- Now, though, for years the cost of storage and of accessing data from storage has been far lower than the cost of having humans actually manage and "keep clean" all that data.
- Recognising that large global businesses tend to have access to new technology first, why is it, then, that few of the largest mature corporations have really shown giant strides with AI?
- One key reason is, yes, data. I know of several that realised several years ago that to fully benefit from AI, they would now have to go through the exercise of cleaning up their data, and when they started to really look at it, they found that they had a LOT of "version control" to do!
- So, the "down and dirty" work of data clean-up has been underway for large corporates for some time, and for mid-size and SMEs some are only just now recognising the drag this will place on their ability to innovate their business with AI.
- In the meantime, we can notice how much "AI middleware" is emerging, promising that it can find data, patterns, and knowledge from within our "data swamps" (as opposed to data lakes). This is very tempting, given it is so difficult to clean up those swamps.
What is the (economic) endgame. Growth or just M&A?
- We are told today that Anthropic may value themselves at $1.5tn for their IPO.
- That is one big number, but as valuation is always (some fundamentals never change) based on market estimates of future profits, what does that mean in terms of revenue and profit growth that will have to be achieved by Anthropic for IPO buyers to ever make any return?
- Well, Noah Smith has published an excellent thought piece on it. In short, Anthropic can't possibly generate real productivity based growth to justify their valuation, they will have to do it by using competitive advantage to gobble up existing companies with their capital. In short, an M&A spree beyond anything we have ever seen.
- https://substack.com/@noahpinion/p-223102188
Was Keynes right about the three-hour workday?
- Nearly 100 years ago (1930), Keynes forecast that increasing economic productivity would have increased so much (4-8x) within the next 100 years (it has increased 5.5x) that a real problem for society would be how we meaningfully fill our time, so the workweek would be cut to around 15 hours so that we could spread out the work to be done enough so that we could all find meaning in that.
- Despite 5.5x productivity growth, many are working longer and harder than ever before, though the developed world seems not to want to acknowledge that, along with social inequality, we have increasingly large swathes of society who are doing no work at all.
- For those who have the privilege and/or choice to work as much as they want, not as much as they have to, meaning in life can absolutely come (at least in part) from work and that is why some keep doing at least some work (I count myself in this number), but at the same time there are increasingly large numbers who want/need to work but can't find work. This may well increase due to AI, leading not just to a loss of meaning in life, but, more directly, to social unrest (as history shows us happens as inequality grows to extremes).
- Anyone reading this can tell that I do look at a lot of numbers and behaviour through the eyes of an economist, so for those on the far end of the free-market side (eg Jacob Rees-Mogg recently declaring that he does not believe in redistribution of income or wealth at any level at all, people should work for a living, etc.), do we want to live as the wealthier do in the most unequal societies, with social unrest meaning we have to live behind fences and gates with security stopping us from having our bubble popped?
Thanks again, Neil, your conversation starter got me thinking.
I know my views are on the more alarmist side of things, but I honestly believe that within a few months AI may be powerful enough to cause human extinction, and that society should prioritise thinking about that and making sure it doesn't happen, before worrying about economic effects. To me "jobs, inequality etc." are long term issues we can worry about later and preventing a catastrophe that might wipe out civilisation, or humanity, is the short term issue we should worry about right now.
Agreed, eg Dario Amodei's essays etc where even he talks about a 15% chance of this. Sometimes reading science fiction (or watching science fiction movies) is scarily prescient too