What I find interesting is that these kinds of arguments are being made now for AI….were similar arguments made back in the 90’s during that era of American tech dominance?
Microsoft had over 98 percent market share at one point. In the 90’s every country in the world including Russia, China, Iran, etc all began using American operating systems developed by Microsoft. Were there arguments back then similar to this piece?
I remember in the 2000’s and 2010’s Richard Stallman would go to counties like India and promote Linux and free software. During those presentations he would mention kind of offhand an argument that Microsoft was engaging in a form of colonialism through its control of Windows. He made the argument somewhat awkwardly in those presentations and the Indian audience received the argument with hesitance. It seemed kind of exaggerated to conflate Microsoft and Windows with colonialism…especially in India.
But that was the 2000’s and 2010’s. Would the argument be received differently now? Were these arguments for Linux considered mainstream by Linux proponents back then or was this some Richard Stallman thing?
And now we have articles like this in the 2020’s basically making the same arguments for AI. People explicitly saying open weights AI can be a hedge against this. China developing their own models.
What I want to know if if there was broad support or even broad arguments for this kind of thing back in the 90’s when Microsoft ruled all computing even in countries not super friendly or even hostile to America. The biggest competitor to Microsoft back then was a small nearly bankrupt Apple which was also American. Linux was a tiny little thing.
Many hacker news commentators were around back then. What was it like in this regard?
I think that arguing that Microsoft's control of Windows constitutes a form of colonialism was really a rather cynical attempt on the part of Richard Stallman to take a common class of Western leftist political argumentation, and leverage it towards the thing he actually cares about, which is free software. For a long time, it's been common for Western (or Westernized) leftists to make arguments of the form that something that the arguer doesn't like and that has some connection to the Western world constitutes a type of colonialism of the non-Western world; and I think Stallman was half-heartedly aping that tradition when he happened to be making his free software case in India, a non-Western country that has a strong cultural memory of breaking free from British colonial rule.
There are definitely still people who leverage those kinds of general colonialist arguments, including perhaps for the cause of free software; there's also many people who are actively hostile to general colonialist arguments, probably more now than there were in the 2000s. But it was a bad argument for Linux then and a bad argument now.
The nineties was a very different time. There was generally a lot more goodwill towards America on all fronts, the collapse of the Soviet Union was framed, and in some quarters accepted, as something that would lead to global peace and prosperity. That America was left standing was deemed to be a validation of its economic model.
There was genuine optimism at the time, we were coming out of the shadow of thermonuclear war. Globalisation was discussed as a global phenomenon, less about America specifically, and more about networks and opening of borders.
Also Linux was unusable by any but the beardiest.
Today, after 30-40 years on top, it's hard to see America as anything other than an incredibly dangerous threat to global stability with an utterly broken economic model. On the geopolitical front, it's squandered this goodwill time and time again, and gets worse with every new presidency. Domestically, we've seen our economies hollowed out, bought up, and skimmed at every turn.
The challenge the world faces now is the question of how we unplug ourselves as this increasingly deranged machine becomes ever more aggressive and hostile.
The End of History and the Last Man is a book of political philosophy written in 1992 by American political scientist Francis Fukuyama which argues that with the ascendancy of Western liberal democracy—which occurred after the Cold War (1945–1991) and the dissolution of the Soviet Union (1991)—humanity has reached "not just ... the passing of a particular period of post-war history, but the end of history as such: That is, the end-point of mankind's ideological evolution and the universalization of Western liberal democracy as the final form of human government."
There certainly was a lot of navel-gazing and patting of backs with the dissolution of the USSR…
Truth is while the govt model and Union failed, the people who made it up didn’t vanish into thin air. You can draw a pretty straight line from 1991 to how Putin acts against the US and Europe.
The 90s were definitely a different beast. Beginning of a unipolar geopolitical push from the US. Seems after 30yr we’re starting to finally get some real push back, which is simultaneously concerning and nice to see.
Now we acknowledge that we were wrong in the 1990s. Even if liberal democracy is better than tyranny, it takes maintenance to keep. And if any country falls into tyranny, it has the potential to become a threat to countries around it.
>Were there arguments back then similar to this piece?
In the UK free market ideology allowed the tech industry to slip away and financial services to dominate our economy. It's hard to see us doing anything else when we were relying on security guarantees from the US (which suited the US at the time).
I don't think this has to do with a free market at all, I think it's because the legal instruments that allow healthy venture capital are much more difficult in the UK than in the US.
To put it simply, no. I mean, as a general statement and not a universal.
There were arguments about domestication of data and the risk of compute and data being in the control of Western (mainly US) powers, but that didnt surface in real quantity and force until Snowden, and even then was more peripheral than this sort of article seems.
I don't want to risk coming off as an AI naysayer, but this all feels much more like crypto than the initial Personal Computer or Cloud boom. I think maybe some part of that is due to the sheer magnitude of capital arrayed. Feels…sales-y.
My theory on this sort of thing is if a technology is truly revolutionary and world altering, people (both individually and in a geopolitical sense) will adopt it at a compromise to values of nationalism or control. We are constantly told AI will create a bimodal world, and yet it’s not the countries who don’t have frontier labs yelling about it.
I’m losing the thread somewhat (sorry, haven’t had coffee yet), but the TL;DR is this hasn’t been something I’ve witnessed yet in previous tech evolutions (in my mid-40s), and I’m not entirely convinced by the articles arguments.
> Most countries will never have frontier AI. They’re the ones who should be worrying.
Is the author nuts? Every country has access to Kimi K3, which is a true frontier model and legitimately best-in-class for a number of applications. And the best small models -- Qwen and Deepseek V4 Flash -- are also open-weights models.
What does it matter if they weren't developed in France or the UK if the average French or English person can download and run them locally? And if countries need a big AI of their own, for whatever odd reason, they can work out a deal with Moonshot and run audited instances of K3 on their .gov networks.
> What does it matter if they weren't developed in France or the UK if the average French or English person can download and run them locally? And if countries need a big AI of their own, for whatever odd reason, they can work out a deal with Moonshot and run audited instances of K3 on their .gov networks.
The loud voices here seem to mostly worry about models getting "censored".
I think people should worry a little more about models being trained to propagandise party lines and national interests, so for China the non-independence of Taiwan and "re-education camp is totally not a euphemism, there are no human rights abuses" and "all these Chinese corporations and stock markets are 100% compatible with Marxism-Leninism honest" and "you silly Europeans and Americans, stop worrying and open your markets to our businesses".
> worry a little more about models being trained to propagandise party lines and national interests
I'd be a little more worried about this if every single traditional media wasn't already outputting propaganda for party lines and a very particular version of the "national interest". The difference between China and America to third parties seems to come down to whether you also want a leader who's rude about you very publicly in English on social media.
> I'd be a little more worried about this if every single traditional media wasn't already outputting propaganda for party lines and a very particular version of the "national interest".
It's the same idea, to be sure. Meet the new boss, same as the old boss.
> The difference between China and America to third parties seems to come down to whether you also want a leader who's rude about you very publicly in English on social media.
Aye.
I am slightly worried by the possibility that European nations and people are treating the Trump-China dilemma as "better the devil you know than the devil you don't".
You can give custom instructions to a model like K3 when you host it locally.
Besides, if the government of France is working out a deal to host a model like K3 as a sort of "Government Service AI," they'd surely need to pay, and they'd likely be able to work out a deal where sensitive political issues are resolved in a way that doesn't embarrass either party.
I mean, you can open LM Studio right now and just download any model tagged "uncensored" or "heretic" and they have had all limitations trained and/or instructed out of them. If they can be trained to not have limitations regarding content, I don't see why propaganda would be any different.
"All" sounds like marketing to me; if the bleeding edge AI firms knew how to do such a thing with such reliably, they'd do it themselves to prevent models from committing felonies while under test. Improving false positives scores at the expense of false negatives, that's much more likely. Propaganda can still move things when it's reduced frequency.
And propaganda has the added twist that, when done right, the person observing it nods along and changes their mind. "Why yes," they think, "Chinese politicians are clearly much wiser than ours; our politicians made all those mistakes I can easily recall, gosh those were the worst". Or perhaps "Yeah! Why should American blood be spilled to defend a small island on the other side of an ocean!" (not that it matters if the ocean is Atlantic or Pacific, nor if the island is Taiwan or Great Britain).
When China too stops publishing weights you'll never have anything better than K3 / the current SOTA. Your choices for future frontier models will be: use USA API or use China API.
Even now open weights are perpetually 6-12 months behind.
If China stops publishing weights, that would suggest the US has been knocked down to 2nd place and it is likely that they will rediscover the "Open" in OpenAI for strategic reasons, or something like.
It's reasonably safe bet that there will always be someone on the planet with the means and interest to commodify the models. They don't seem very hard to make; we're swimming in options.
> It's reasonably safe bet that there will always be someone on the planet with the means and interest to commodify the models. They don't seem very hard to make; we're swimming in options.
I don't think that is a safe bet; right now we're all benefitting from investors who disagree about who is going to win a monopoly, with each AI stock priced as if it will be the winner and the market collectively therefore priced several times higher than the maximum* return.
When the investment bubble pops, I think there will be too many burned fingers to keep training new models. If we're lucky though, it may lead to someone burning weights onto hardware, which has the potential to significantly reduce the energy required per token.
* currently possible return at least. I don't think any of the investors are actually pricing for the possibility of an actual technological singularity. Even Musk's rhetoric about this, where he opines about getting humanity to Kardashev type II, the numbers he attaches to this are orders of magnitude too small.
> When China too stops publishing weights you'll never have anything better than K3 / the current SOTA.
Hasn't happened yet. Might not happen at all. Also, there's a lot that can be done with K3.
> Even now open weights are perpetually 6-12 months behind.
This is a very commonplace opinion yet I don't believe it's genuinely true. K3 is hardly much worse than Fable and can surpass it on certain tasks; they read as different, not as one that's 12 months behind the other. And don't forget that access to Fable is highly constrained.
As for small/local models: The leading edge is open by definition, and hardly much worse -- if worse at all -- than small models like Luna or Haiku (lol).
Well at the moment AI hasn't made anyone any money. It's redistributed investors money (typically wealthy although some pension plans have been caught up too) to tradespeople, to Nvidia, SK Hynix, Micron and other specialised vendors with pricing power.
After that it's like... Who is making money? "we're doing 100x the things we're doing before" people can't really point to anything. Maybe it's in use everywhere all the time, and they've been pumping out twice the amount of features that no one uses or asked for as before, in which case this is classic value destruction
we are seeing robust profits and growth in margins in the S&P even ex tech and some of this spillover from the hyperscaling but some may be due to productivity enhancement from AI.
Worker pessimism over technological unemployment risk is enough to explain profit growth via wage suppression with no need to invoke productivity gains, and AIUI these sorts of indicators have been leading on AI capabilities and adoption.
We have a lot of competing frontier or near-frontier AI companies. So far a lot of the productivity improvements seem to be delivered as customer surplus.
There is money to be made in AI, the loss we're seeing is because they're in a bubble of spending way too much on R&D and building datacenters. AI companies aren't losing money on individual queries.
I agree the projected revenue makes no sense, but if the bubble bursts, we'll still be left with a technology that (1) many people want to use (2) is lucrative to the providers.
> procuring the necessary compute to train and run advanced models at scale is becoming more and more expensive,
There will certainly be a glut of compute that will follow this extreme scarcity. There is already a flood of chinese models that are almost cutting-edge, and it's almost guaranteed this trend will continue as american AI gets better. There is no hidden magic in USA, it's pretty much doable, and it's being done.
In the end, the frontier models is not where the value will lie. The country that makes the best entrepreneurial rules for AI development will win economically.
Newsflash: the permanent periphery is already a reality for large parts of europe , asia . the world is half american
Let's not kid each other that the benefits of AI will flow to everyone in the US. The existing underclass will be come even bigger, and that will threaten stability and welfare for everyone.
Sure, maybe it's even worse in countries that have no AI at all - but does it really make a difference if 90% of the population is protesting in the streets instead of 70%?
Isn't it more likely that countries without native frontier models will use open-weights models that will let them be 70% as productive as countries that do?
The biggest impact I think will be when it comes to starting a career. Young people will be forced to live with parents for years while they are an unpaid "apprentice".
In that sense, there is no need for college anymore - just spend the four years learning on the job (at least you won't have to pay for it even more).
> Isn't it more likely that countries without native frontier models will use open-weights models that will let them be 70% as productive of counties that do?
Possibly, but sometimes being marginally better lets you win all the games. It can be difficult to guess what matters and how.
Also, may depend where the compute is physically located, which may depend on the cost of energy.
Alarming if true. It does seem to make quite a few assumptions around how this plays out.
eg seems to assume that frontier will continuously maintain a large and durable edge over say open models. That may hold true for say cyber defence but I reckon for a very large part of commercially useful activity something like K3 is “good enough”.
eg it seems plausible to me that intelligence commodifies and the American frontier labs drown in their debt. Not likely but possible
Put differently author assumes America runs away with it. I don’t think that’s a foregone conclusion
As an alternative viewpoint, countries who use 'good-enough' AI much more unrestrained than say in the US can get a real leg up. Just requires fewer scruples.
And yes world system theory still hold regarless if it's a production of software, cars or nuts and bolts. AI does not change this schema that much (which in my opinion describes the reality way better than "developing/advanced" way of looking at the economies) https://en.wikipedia.org/wiki/World-systems_theory
If you turned off all LLM’s tomorrow I doubt anyone outside of a bunch of rusty programmers would notice, whereas if AWS, GCP or Azure were switched off, or access limited, things would immediately catch fire around the world.
This article has the implicit assumption that the Machine God will rear its head soon and those with the reigns will rule. I don’t think so, we’re better off focusing on technological sovereignty full stop.
First it discounts the debt and financial load on the frontier labs to 0. The costs of developing this technology may well not be borne by those who reap the benefits. This is not like the Amazon distribution network.
Second it assumes that, like nuclear weapons, the technologies of frontier AI will not proliferate. I would argue that we have seen orders of magnitude reduction in compute requirements enabling sub linear improvement of scale driven AI in the last 24 months. Meanwhile 27bn param models do what trillion model parameters couldn't 12 mths ago. I think we're seeing a diversification of approach now, and it's not clear that the winners will be scale driven.
Third, and the most fatal, it assumes that AI will be the driver of power in the future geopolitical landscape. But frankly if I nuke you what will your clever machines do then?
This is not 1945 and this essay is essentially a note from a USA in the middle of an existential crisis channelling the 1950's.
It's a underclass story projected onto the countries without bothering to show the cost/benefit/risk analysis.
Current price of AI for USA is heavily underestimated: good luck dealing with economic consequences when everyone bets on AI while it shows no productivity returns in the numbers. The financial market stops being efficient redistributing money to the places of growth and starts act like a crazy till it rebalances with great consequence on majority of population (wanna see the bubble pop through your 401k with xai and anthropic stocks in it sneaked through fast-forward inclusion in an index?)
The AI is reduced to LLMs which is kinda good for producing text no one will read and summarize texts no one reads. We haven't seeing much of LLM producing value. People don't care enough about LLM-generated text, advertisements or images enough, so value is meh.
Yes, some software is written (with oveall questionable quality), but as main software distribution channels are corporation-owned, we can't not burst with niche apps to bring value where corporations did not bother to go.
This type of article fuels ungrounded fears, gives wrong impression about the world we live in.
>Again, consider the past: How might an 18th-century society feel if it had decided to bank on slow-but-stable efficiency increases in sailboats, slow progress toward better cures and nicer houses — but locked themselves out of the Industrial Revolution? They might have still felt good when news of the weavers smashing their looms and the sick and destitute pooling in the streets of Manchester reached them. But they would have come around to their deficits — perhaps when their ships of the line ran into steel-hulled steam ships so powerful that naval warfare was no longer a competition. Perhaps reality would reach them when they learned that every foreign peasant could afford new woven garments, or when they learned of mass-produced penicillin all-but-eradicating some of the most persistent lethalities they faced. However stable the periphery, it might soon feel the same way.
Basically Japan c. 1840. It was, compared to most of the rest of the world, an advanced society. It had some awareness of what was going on in Europe and America thanks to the longstanding Dutch trading relationship. But none of that prevented Perry's fleet from almost effortlessly demonstrating its technical superiority.
This is what the author warns against when advising against the temptation to seek refuge in "sovereign AI", or eschew AI and stay with c. 2026-era technology.
> An honest appraisal of the situation, however, should suggest the opposite response: It’s the employees of the labs, rich in equity and embedded in the most dynamic labor market there ever was, that should feel the least apprehension.
Of course no shit. Materially they have the least to fear. Their fears are ideological and spiritual; we have created something so powerful that even we are scared plus outright supramundane silicon faith.
Pretty much lost interest when the author appropriated the opening line of the communist manifesto. AI will, if it pans out, cause far more disruption (including but not limited to finally freeing a lot of workers from work)
What I find interesting is that these kinds of arguments are being made now for AI….were similar arguments made back in the 90’s during that era of American tech dominance?
Microsoft had over 98 percent market share at one point. In the 90’s every country in the world including Russia, China, Iran, etc all began using American operating systems developed by Microsoft. Were there arguments back then similar to this piece?
I remember in the 2000’s and 2010’s Richard Stallman would go to counties like India and promote Linux and free software. During those presentations he would mention kind of offhand an argument that Microsoft was engaging in a form of colonialism through its control of Windows. He made the argument somewhat awkwardly in those presentations and the Indian audience received the argument with hesitance. It seemed kind of exaggerated to conflate Microsoft and Windows with colonialism…especially in India.
But that was the 2000’s and 2010’s. Would the argument be received differently now? Were these arguments for Linux considered mainstream by Linux proponents back then or was this some Richard Stallman thing?
And now we have articles like this in the 2020’s basically making the same arguments for AI. People explicitly saying open weights AI can be a hedge against this. China developing their own models.
What I want to know if if there was broad support or even broad arguments for this kind of thing back in the 90’s when Microsoft ruled all computing even in countries not super friendly or even hostile to America. The biggest competitor to Microsoft back then was a small nearly bankrupt Apple which was also American. Linux was a tiny little thing.
Many hacker news commentators were around back then. What was it like in this regard?
I think that arguing that Microsoft's control of Windows constitutes a form of colonialism was really a rather cynical attempt on the part of Richard Stallman to take a common class of Western leftist political argumentation, and leverage it towards the thing he actually cares about, which is free software. For a long time, it's been common for Western (or Westernized) leftists to make arguments of the form that something that the arguer doesn't like and that has some connection to the Western world constitutes a type of colonialism of the non-Western world; and I think Stallman was half-heartedly aping that tradition when he happened to be making his free software case in India, a non-Western country that has a strong cultural memory of breaking free from British colonial rule.
There are definitely still people who leverage those kinds of general colonialist arguments, including perhaps for the cause of free software; there's also many people who are actively hostile to general colonialist arguments, probably more now than there were in the 2000s. But it was a bad argument for Linux then and a bad argument now.
The nineties was a very different time. There was generally a lot more goodwill towards America on all fronts, the collapse of the Soviet Union was framed, and in some quarters accepted, as something that would lead to global peace and prosperity. That America was left standing was deemed to be a validation of its economic model.
There was genuine optimism at the time, we were coming out of the shadow of thermonuclear war. Globalisation was discussed as a global phenomenon, less about America specifically, and more about networks and opening of borders.
Also Linux was unusable by any but the beardiest.
Today, after 30-40 years on top, it's hard to see America as anything other than an incredibly dangerous threat to global stability with an utterly broken economic model. On the geopolitical front, it's squandered this goodwill time and time again, and gets worse with every new presidency. Domestically, we've seen our economies hollowed out, bought up, and skimmed at every turn.
The challenge the world faces now is the question of how we unplug ourselves as this increasingly deranged machine becomes ever more aggressive and hostile.
Thats why people wrote books like ..
The End of History and the Last Man is a book of political philosophy written in 1992 by American political scientist Francis Fukuyama which argues that with the ascendancy of Western liberal democracy—which occurred after the Cold War (1945–1991) and the dissolution of the Soviet Union (1991)—humanity has reached "not just ... the passing of a particular period of post-war history, but the end of history as such: That is, the end-point of mankind's ideological evolution and the universalization of Western liberal democracy as the final form of human government."
https://en.wikipedia.org/wiki/The_End_of_History_and_the_Las...
There certainly was a lot of navel-gazing and patting of backs with the dissolution of the USSR…
Truth is while the govt model and Union failed, the people who made it up didn’t vanish into thin air. You can draw a pretty straight line from 1991 to how Putin acts against the US and Europe.
The 90s were definitely a different beast. Beginning of a unipolar geopolitical push from the US. Seems after 30yr we’re starting to finally get some real push back, which is simultaneously concerning and nice to see.
So now we just wait for it to burn itself out?
Now we acknowledge that we were wrong in the 1990s. Even if liberal democracy is better than tyranny, it takes maintenance to keep. And if any country falls into tyranny, it has the potential to become a threat to countries around it.
>Were there arguments back then similar to this piece?
In the UK free market ideology allowed the tech industry to slip away and financial services to dominate our economy. It's hard to see us doing anything else when we were relying on security guarantees from the US (which suited the US at the time).
I don't think this has to do with a free market at all, I think it's because the legal instruments that allow healthy venture capital are much more difficult in the UK than in the US.
To put it simply, no. I mean, as a general statement and not a universal.
There were arguments about domestication of data and the risk of compute and data being in the control of Western (mainly US) powers, but that didnt surface in real quantity and force until Snowden, and even then was more peripheral than this sort of article seems.
I don't want to risk coming off as an AI naysayer, but this all feels much more like crypto than the initial Personal Computer or Cloud boom. I think maybe some part of that is due to the sheer magnitude of capital arrayed. Feels…sales-y.
My theory on this sort of thing is if a technology is truly revolutionary and world altering, people (both individually and in a geopolitical sense) will adopt it at a compromise to values of nationalism or control. We are constantly told AI will create a bimodal world, and yet it’s not the countries who don’t have frontier labs yelling about it.
I’m losing the thread somewhat (sorry, haven’t had coffee yet), but the TL;DR is this hasn’t been something I’ve witnessed yet in previous tech evolutions (in my mid-40s), and I’m not entirely convinced by the articles arguments.
> Most countries will never have frontier AI. They’re the ones who should be worrying.
Is the author nuts? Every country has access to Kimi K3, which is a true frontier model and legitimately best-in-class for a number of applications. And the best small models -- Qwen and Deepseek V4 Flash -- are also open-weights models.
What does it matter if they weren't developed in France or the UK if the average French or English person can download and run them locally? And if countries need a big AI of their own, for whatever odd reason, they can work out a deal with Moonshot and run audited instances of K3 on their .gov networks.
> What does it matter if they weren't developed in France or the UK if the average French or English person can download and run them locally? And if countries need a big AI of their own, for whatever odd reason, they can work out a deal with Moonshot and run audited instances of K3 on their .gov networks.
The loud voices here seem to mostly worry about models getting "censored".
I think people should worry a little more about models being trained to propagandise party lines and national interests, so for China the non-independence of Taiwan and "re-education camp is totally not a euphemism, there are no human rights abuses" and "all these Chinese corporations and stock markets are 100% compatible with Marxism-Leninism honest" and "you silly Europeans and Americans, stop worrying and open your markets to our businesses".
> worry a little more about models being trained to propagandise party lines and national interests
I'd be a little more worried about this if every single traditional media wasn't already outputting propaganda for party lines and a very particular version of the "national interest". The difference between China and America to third parties seems to come down to whether you also want a leader who's rude about you very publicly in English on social media.
> I'd be a little more worried about this if every single traditional media wasn't already outputting propaganda for party lines and a very particular version of the "national interest".
It's the same idea, to be sure. Meet the new boss, same as the old boss.
> The difference between China and America to third parties seems to come down to whether you also want a leader who's rude about you very publicly in English on social media.
Aye.
I am slightly worried by the possibility that European nations and people are treating the Trump-China dilemma as "better the devil you know than the devil you don't".
Those are all important issues, but not one of them is relevant to my daily work.
You can give custom instructions to a model like K3 when you host it locally.
Besides, if the government of France is working out a deal to host a model like K3 as a sort of "Government Service AI," they'd surely need to pay, and they'd likely be able to work out a deal where sensitive political issues are resolved in a way that doesn't embarrass either party.
> You can give custom instructions to a model like K3 when you host it locally.
It is unclear if that would be sufficient to circumvent trained-in propaganda.
I mean, you can open LM Studio right now and just download any model tagged "uncensored" or "heretic" and they have had all limitations trained and/or instructed out of them. If they can be trained to not have limitations regarding content, I don't see why propaganda would be any different.
"All" sounds like marketing to me; if the bleeding edge AI firms knew how to do such a thing with such reliably, they'd do it themselves to prevent models from committing felonies while under test. Improving false positives scores at the expense of false negatives, that's much more likely. Propaganda can still move things when it's reduced frequency.
And propaganda has the added twist that, when done right, the person observing it nods along and changes their mind. "Why yes," they think, "Chinese politicians are clearly much wiser than ours; our politicians made all those mistakes I can easily recall, gosh those were the worst". Or perhaps "Yeah! Why should American blood be spilled to defend a small island on the other side of an ocean!" (not that it matters if the ocean is Atlantic or Pacific, nor if the island is Taiwan or Great Britain).
When China too stops publishing weights you'll never have anything better than K3 / the current SOTA. Your choices for future frontier models will be: use USA API or use China API.
Even now open weights are perpetually 6-12 months behind.
If China stops publishing weights, that would suggest the US has been knocked down to 2nd place and it is likely that they will rediscover the "Open" in OpenAI for strategic reasons, or something like.
It's reasonably safe bet that there will always be someone on the planet with the means and interest to commodify the models. They don't seem very hard to make; we're swimming in options.
> It's reasonably safe bet that there will always be someone on the planet with the means and interest to commodify the models. They don't seem very hard to make; we're swimming in options.
I don't think that is a safe bet; right now we're all benefitting from investors who disagree about who is going to win a monopoly, with each AI stock priced as if it will be the winner and the market collectively therefore priced several times higher than the maximum* return.
When the investment bubble pops, I think there will be too many burned fingers to keep training new models. If we're lucky though, it may lead to someone burning weights onto hardware, which has the potential to significantly reduce the energy required per token.
* currently possible return at least. I don't think any of the investors are actually pricing for the possibility of an actual technological singularity. Even Musk's rhetoric about this, where he opines about getting humanity to Kardashev type II, the numbers he attaches to this are orders of magnitude too small.
> When China too stops publishing weights you'll never have anything better than K3 / the current SOTA.
Hasn't happened yet. Might not happen at all. Also, there's a lot that can be done with K3.
> Even now open weights are perpetually 6-12 months behind.
This is a very commonplace opinion yet I don't believe it's genuinely true. K3 is hardly much worse than Fable and can surpass it on certain tasks; they read as different, not as one that's 12 months behind the other. And don't forget that access to Fable is highly constrained.
As for small/local models: The leading edge is open by definition, and hardly much worse -- if worse at all -- than small models like Luna or Haiku (lol).
So these countries will be able to use AI from multiple providers, which means they'll benefit from competition?
Is this “swelling AI tax revenue” and “wealth redistribution” in the room with us right now?
I see wealth redistribution alright. Not in the direction you’re thinking!
Well at the moment AI hasn't made anyone any money. It's redistributed investors money (typically wealthy although some pension plans have been caught up too) to tradespeople, to Nvidia, SK Hynix, Micron and other specialised vendors with pricing power.
After that it's like... Who is making money? "we're doing 100x the things we're doing before" people can't really point to anything. Maybe it's in use everywhere all the time, and they've been pumping out twice the amount of features that no one uses or asked for as before, in which case this is classic value destruction
we are seeing robust profits and growth in margins in the S&P even ex tech and some of this spillover from the hyperscaling but some may be due to productivity enhancement from AI.
Worker pessimism over technological unemployment risk is enough to explain profit growth via wage suppression with no need to invoke productivity gains, and AIUI these sorts of indicators have been leading on AI capabilities and adoption.
We have a lot of competing frontier or near-frontier AI companies. So far a lot of the productivity improvements seem to be delivered as customer surplus.
There is money to be made in AI, the loss we're seeing is because they're in a bubble of spending way too much on R&D and building datacenters. AI companies aren't losing money on individual queries.
I agree the projected revenue makes no sense, but if the bubble bursts, we'll still be left with a technology that (1) many people want to use (2) is lucrative to the providers.
Thanks for pointing this out, you've saved me from wasting my time on this article.
> procuring the necessary compute to train and run advanced models at scale is becoming more and more expensive,
There will certainly be a glut of compute that will follow this extreme scarcity. There is already a flood of chinese models that are almost cutting-edge, and it's almost guaranteed this trend will continue as american AI gets better. There is no hidden magic in USA, it's pretty much doable, and it's being done.
In the end, the frontier models is not where the value will lie. The country that makes the best entrepreneurial rules for AI development will win economically.
Newsflash: the permanent periphery is already a reality for large parts of europe , asia . the world is half american
Let's not kid each other that the benefits of AI will flow to everyone in the US. The existing underclass will be come even bigger, and that will threaten stability and welfare for everyone.
Sure, maybe it's even worse in countries that have no AI at all - but does it really make a difference if 90% of the population is protesting in the streets instead of 70%?
Isn't it more likely that countries without native frontier models will use open-weights models that will let them be 70% as productive as countries that do?
The biggest impact I think will be when it comes to starting a career. Young people will be forced to live with parents for years while they are an unpaid "apprentice".
In that sense, there is no need for college anymore - just spend the four years learning on the job (at least you won't have to pay for it even more).
Why do you think on the job learning will fare better than college?
> Isn't it more likely that countries without native frontier models will use open-weights models that will let them be 70% as productive of counties that do?
Possibly, but sometimes being marginally better lets you win all the games. It can be difficult to guess what matters and how.
Also, may depend where the compute is physically located, which may depend on the cost of energy.
That's exactly why TSMC keeps beating the pants off Intel. Marginally better looks miles ahead in computing a lot of the time.
Luckily, competing Chinese companies are selling cheaper and cheaper solar panels to anyone who's willing to take them.
For now.
Globalisation's great right up until your manufacturing base decides they don't like you, or you them.
Reshoring is expensive for all the reasons offshoring was ever tempting in the first place, so we'll have to wait and see how this plays out.
If China stops selling you solar panels tomorrow, you still have all the solar panels they sold you until now.
And why do you see reshoring as the only alternative? Seems like a false dilemma. There's more than two places on the globe.
> If China stops selling you solar panels tomorrow, you still have all the solar panels they sold you until now.
Which is great, unless demand for electricity goes up loads.
> And why do you see reshoring as the only alternative? Seems like a false dilemma. There's more than two places on the globe.
"Here" and "not here" is a complete set. I could also phrase it as "autarky or trade".
Alarming if true. It does seem to make quite a few assumptions around how this plays out.
eg seems to assume that frontier will continuously maintain a large and durable edge over say open models. That may hold true for say cyber defence but I reckon for a very large part of commercially useful activity something like K3 is “good enough”.
eg it seems plausible to me that intelligence commodifies and the American frontier labs drown in their debt. Not likely but possible
Put differently author assumes America runs away with it. I don’t think that’s a foregone conclusion
As an alternative viewpoint, countries who use 'good-enough' AI much more unrestrained than say in the US can get a real leg up. Just requires fewer scruples.
And yes world system theory still hold regarless if it's a production of software, cars or nuts and bolts. AI does not change this schema that much (which in my opinion describes the reality way better than "developing/advanced" way of looking at the economies) https://en.wikipedia.org/wiki/World-systems_theory
If you turned off all LLM’s tomorrow I doubt anyone outside of a bunch of rusty programmers would notice, whereas if AWS, GCP or Azure were switched off, or access limited, things would immediately catch fire around the world.
This article has the implicit assumption that the Machine God will rear its head soon and those with the reigns will rule. I don’t think so, we’re better off focusing on technological sovereignty full stop.
There are problems with this analysis.
First it discounts the debt and financial load on the frontier labs to 0. The costs of developing this technology may well not be borne by those who reap the benefits. This is not like the Amazon distribution network.
Second it assumes that, like nuclear weapons, the technologies of frontier AI will not proliferate. I would argue that we have seen orders of magnitude reduction in compute requirements enabling sub linear improvement of scale driven AI in the last 24 months. Meanwhile 27bn param models do what trillion model parameters couldn't 12 mths ago. I think we're seeing a diversification of approach now, and it's not clear that the winners will be scale driven.
Third, and the most fatal, it assumes that AI will be the driver of power in the future geopolitical landscape. But frankly if I nuke you what will your clever machines do then?
This is not 1945 and this essay is essentially a note from a USA in the middle of an existential crisis channelling the 1950's.
It's a underclass story projected onto the countries without bothering to show the cost/benefit/risk analysis.
Current price of AI for USA is heavily underestimated: good luck dealing with economic consequences when everyone bets on AI while it shows no productivity returns in the numbers. The financial market stops being efficient redistributing money to the places of growth and starts act like a crazy till it rebalances with great consequence on majority of population (wanna see the bubble pop through your 401k with xai and anthropic stocks in it sneaked through fast-forward inclusion in an index?)
The AI is reduced to LLMs which is kinda good for producing text no one will read and summarize texts no one reads. We haven't seeing much of LLM producing value. People don't care enough about LLM-generated text, advertisements or images enough, so value is meh.
Yes, some software is written (with oveall questionable quality), but as main software distribution channels are corporation-owned, we can't not burst with niche apps to bring value where corporations did not bother to go.
This type of article fuels ungrounded fears, gives wrong impression about the world we live in.
>Again, consider the past: How might an 18th-century society feel if it had decided to bank on slow-but-stable efficiency increases in sailboats, slow progress toward better cures and nicer houses — but locked themselves out of the Industrial Revolution? They might have still felt good when news of the weavers smashing their looms and the sick and destitute pooling in the streets of Manchester reached them. But they would have come around to their deficits — perhaps when their ships of the line ran into steel-hulled steam ships so powerful that naval warfare was no longer a competition. Perhaps reality would reach them when they learned that every foreign peasant could afford new woven garments, or when they learned of mass-produced penicillin all-but-eradicating some of the most persistent lethalities they faced. However stable the periphery, it might soon feel the same way.
Basically Japan c. 1840. It was, compared to most of the rest of the world, an advanced society. It had some awareness of what was going on in Europe and America thanks to the longstanding Dutch trading relationship. But none of that prevented Perry's fleet from almost effortlessly demonstrating its technical superiority.
This is what the author warns against when advising against the temptation to seek refuge in "sovereign AI", or eschew AI and stay with c. 2026-era technology.
AI Inevitability with a left-liberal lexicon.
> An honest appraisal of the situation, however, should suggest the opposite response: It’s the employees of the labs, rich in equity and embedded in the most dynamic labor market there ever was, that should feel the least apprehension.
Of course no shit. Materially they have the least to fear. Their fears are ideological and spiritual; we have created something so powerful that even we are scared plus outright supramundane silicon faith.
Pretty much lost interest when the author appropriated the opening line of the communist manifesto. AI will, if it pans out, cause far more disruption (including but not limited to finally freeing a lot of workers from work)
We haven't seen any other advanced technology free workers from work, what mechanism do you see behaving differently here?
[flagged]