> 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.
> 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.
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.
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?
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.
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)
> 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
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?
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.
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.
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
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
>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.
16 comments
[ 1.1 ms ] story [ 25.9 ms ] thread> 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.
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.
Even a month can change a lot these times.
I see wealth redistribution alright. Not in the direction you’re thinking!
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?
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.
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
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?
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.
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.
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
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.