On Jacob Coxon s Resignation, AI, and the Extinction of Humanity

Since the “end of history” didn’t work out and the “end of work” — at least for now — hasn’t either, we now have the “end of humanity.” 

Allow me to be skeptical. I have the feeling the greatest danger stems from the capitalist implementation of AI on a large scale, combined with other technologies — and their many social, political, and economic consequences — rather than from AI itself. This implementation, if fully realized, would entail both a catastrophic defeat and a geopolitical realignment on a scale comparable, at the very least, to that of neoliberalism. Just as it was back then, these are processes that, in a way, are unfolding in stages, through existing levels of technological application, wars, and uncontrolled investment and debt in limited sectors such as AI in the U.S. and its potential ramifications.

This is not to say that the development of AI and algorithmic agents in and of itself does not entail risks, which — in my opinion and, for the time being — are unlikely to pose a danger comparable to that of, for example, atomic energy and the possibility of the planet’s total destruction. The question of AI’s “total autonomy” can never be ruled out in theory, but it remains in the realm of pure speculation, and nothing of the sort exists today. Not even Artificial General Intelligence — which the “Magnificent Seven” are betting on based on the logic of scaling — is a reality. It is still a long way off, and there are even doubts as to whether it can ever be achieved, let alone superintelligence, which would represent true machine autonomy and is considered by a large number of specialists and scientists to be a kind of delusion.

In this context, it is worth highlighting a few points.

1) This entire discussion about the catastrophic risks of AI brings to mind, once again, the famous quote attributed to Fredric Jameson that it is easier to imagine the end of the world than the end of capitalism. If we look back a bit, dystopian prophecies recur and generally stem from a valid point, like ecological collapse. But while we all amuse — and terrify — ourselves with these prophecies, what is creating an increasingly dystopian world is the madness of capital’s economic logic, which is becoming ever more pathological. This is a logic that few seem to fear.

2) There is a runaway campaign by the companies themselves — nine of the ten most valuable companies in the world are tech firms based in the U.S. — as well as by their engineers and CEOs, who either remain with these companies or resign, often moving on to another company, warning that the product they sell is extremely dangerous… which is strange. Many quite credible voices consider this a true historical “singularity.” 

This is a fear campaign that appears linked to the accumulation of power: no company or government wants to be left out of something potentially so risky, even if it doesn’t generate significant profits. Part of the campaign is the idea that if they don’t keep investing trillions in scaling — that is, building more data centers with more cutting-edge AI chips, equipment, and frenzied energy and water consumption, not to mention environmental pollution — China will win the race. 

All of this is fueling sky-high stock prices, meaning it’s allowing them to rack up extraordinary profits. A fairly reputable U.S. nonprofit has been saying for several months now that this corporate rhetoric about “beating China” is laying the groundwork for a bailout if the investment bubble bursts, and that the government is already developing rescue mechanisms. 

There is a very large discrepancy between investment in AI and other advanced technology in the U.S. and the rest of the economy (year-over-year investment growth is negative in low-tech sectors), which poses two risks. 

On the one hand, there is the possibility that the investment bubble will burst (AI itself is not a bubble, of course, but a major financial bubble may be forming around it). On the other hand, there is the risk of crowding out — that AI will end up swallowing up the rest of the economy, which is unable to take off, and causing widespread stagnation. Obviously, it is impossible to know what will happen, but when capital moves so much money amid such structural disparities, the risk is high and the rhetoric is unreliable.

3) Tech leaders have already promised utopias and dystopias of every kind: the end of history and of work, as envisioned in the 1990s; the universal cure for all diseases when 90 percent of the human genome was sequenced (these hopes collapsed along with the dot-com bubble in 2001); stem cells, which were supposed to cure everything; robots that were going to replace all jobs (the second promise of the end of work); cryptocurrencies that were going to dominate the entire economy; the “Metaverse” in which we were supposed to have been living for several years now, and on and on.

4) While companies spread mass fear about the lack of control over their own products, which they continue to develop without limits, the more immediate reality is the control they are achieving over workers and the population at large through the large-scale application of AI: in everyday life; in commerce, where AI can help move products made obsolete through planned obsolescence by selling any kind of junk; and in finance, through countless applications that are drowning everyone in debt. The most dramatic specific example of “loss of control” so far was the attack on the girls’ school in Iran, which may have involved Anthropic’s Claude AI software; however, this type of “lethal error” is not new in military history and has previously been carried out using natural intelligence.

5) Finally, on the issue of AI regulations: I do not believe, in principle, that one can support or reject en bloc any idea of regulation, which in and of itself cannot be defined as progressive or regressive. In fact, the global debate is framed as “government regulation” — yes or no — Anthropic vs. Palantir, the Pope (encyclical Magnifica Humanitas) vs. unbridled (neo)liberalism, or the United States vs. the European Union. 

And, as always, the problem is that the content of government regulation mixes defensible aspects with highly reactionary ones. Regulation is not neutral, and the areas where it could be applied are practically endless. For example, progressive issues such as giving users some control over algorithms (a project in Australia) are mixed with the ban or restriction in many countries on minors under 16 accessing social media — a measure with which I personally do not agree in principle, especially since it also reinforces surveillance of parents, that is, of almost everyone. 

Another significant issue is the line between reactionary transhumanism and transgender interventions, or the possibility of overcoming physical limitations — a topic very well addressed, for example, by Facundo Nahuel Martín in his book Ilustración sensible. These are just a few examples, but it’s worth keeping them in mind. 

In this highly complex context, I am inclined to support any legitimate process — such as, for example, the public outcry in New York against data centers, which Mayor Zohran Mamdani was forced to ban for a time. And, more generally, I support the idea of highly progressive regulations — such as the rejection of biometric controls both within and outside the workplace, or the rejection of fraudulent financial apps — linked to the discussions, decisions, and oversight of the diverse and numerous sectors of workers and the popular majority. 

This support and solidarity must always be linked to a sharp critique of private property in order to challenge its legitimacy. Although significant popular movements may manage to set certain limits, broad and progressive regulation is unthinkable without questioning the uses of technologies that are, obviously, intimately linked to the aims of those who own them. Within that tension, the balance of power is at stake. 

The question, then, is to what extent capital is capable of advancing through the large-scale application not only of AI, but of qualitatively improved AI fused with qualitatively improved robotics and other technologies, such as biotechnology and nanotechnology, to transform the economy and life. At the same time, it is a question of to what extent the working class and popular sectors are capable of advancing by questioning those very ends — and, therefore, the ownership of these technologies.

Regarding this tension, in the final chapter of my book, Las nuevas tecnologías en los límites del capital, I propose the idea of technological hegemony. This idea emerges as a critique of the currently existing capitalist hegemony and as a territory of dispute linked to the “general purposes” of technologies, in contrast to the project of universal basic income, which I view as anti-hegemonic. 

I propose the construction of a technological hegemony as a necessity for advancing toward a qualitatively different distribution of working time that would simultaneously increase free time and give prominence to the utility of work, the goods and services produced, and nature itself. I present this idea as a concept still under development, engaging in a discussion of the “ethics of work” and prioritizing the areas that I consider key to thinking through its development.

Originally published in Spanish on September 15 in La Izquierda Diario