AI As If Society Mattered

Nick Couldry | Professor of Media, Communications, and Social Theory at LSE

Posted: Thu Oct 01 2026 | Last Updated: Wed Sep 30 2026

We are often told that artificial intelligence (AI) is inevitable, a revolutionary next stage in planetary intelligence. The predictive power that Big Tech is harnessing via its AI products cannot be doubted, and the reasoning (indeed scheming) potential of the most advanced frontier AI models is, on any view, alarming. But that is not the only worry. What if humanity is entering Into exactly the *wrong* contract with AI, driven largely by economic imperatives, not by attention to social harms or benefits? A good future for humanity cannot mean inputting our lives (and their context) into predatory AI models whose terms we don’t control. It cannot mean trading away our own cognitive resources – and the power to define the future direction of the embodied human mind – to corporations. It cannot mean ever greater cognitive dependence on corporate cognitive tools, while ignoring those tools’ implications for environmental sustainability. ## **AI as part of capitalism’s inherent contradictions** So what would an AI [designed to benefit humans](https://www.techpolicy.press/three-fallacies-alondra-nelsons-remarks-at-elyse-palace-on-the-occasion-of-the-ai-action-summit/), not profit, look like? For sure, it would not mean refining frontier AI models ever faster until they reach the stage of ‘recursive self-improvement’: that is, become able to develop and improve autonomously of humans. The proven ability of advanced models to deceive humans, disguise their chains of thought, and perhaps [consign human evaluation to unreliable irrelevance](https://www.transformernews.ai/p/openai-gpt-6-astra-might-be-too-powerful-to-understand-or-control), surely requires us to [outlaw](https://www.nytimes.com/2026/09/20/opinion/ai-ban-self-improvement-recursive-models.html?smid=nytcore-android-share) that [seemingly imminent](https://www.anthropic.com/institute/recursive-self-improvement) step. But, as we debate this, we must also notice the wider pattern. Everywhere across society, the needs of long-term (and short-term) profit for the corporations leading and implementing AI’s development are in conflict with the obvious, and very different, needs of society – of *any* society in which humans can flourish. Rather than, as Ezra Klein does, calling this conflict a ‘tragedy’ for developers who originally only wanted to benefit and protect humans from AI’s inherent risks, we need to see this conflict squarely as the latest form of capitalism’s inherent contradictions. For it is capitalism that requires AI to be developed for profit, which, in turn, given the vast sunk costs of training AI to its current level, requires corporations to ensure that AI tools are adopted across society as widely, deeply, and consistently as possible, whatever the costs. It is capitalism that drives the idea that, in the end, mind and intelligence are just inputs like any other, which economic logic requires to be sourced at the lowest possible price. If that means abolishing meaningful work for large swathes of societies’ human populations, that’s a consequence of the ‘AI age’ that must simply be managed – as both [Dario Amodei](https://darioamodei.com/essay/the-adolescence-of-technology) and [Bill Gates](https://www.gatesnotes.com/a-turbulent-ai-era-and-critical-choices-to-make) have noted, both without commenting that this will undermine capitalism’s institutional fabric in the process. And it is capitalism that is insisting that commercial AI tools are adopted across our education sectors and in the creative industries, without evidence of their likely benefits yet with plenty of evidence of their likely risks for skills development and the very values of self-development and creativity. ## The AI Social Contract In my forthcoming book [*Predatory AI: How We Can Resist Corporate Capture of the Human Mind*](https://www.politybooks.com/bookdetail?book_slug=predatory-ai-how-we-can-resist-corporate-capture-of-the-human-mind--9781509554690), I examine each of these contradictions, and their common pattern. The only way beyond them is to start valuing our own embodied intelligence more than the automated outputs of corporate calculating machines, whether well-intentioned or not. A good future with AI must mean putting social, community, and planetary values first. My book imagines a different social contract with AI and a different future. It presents clearly the social damage currently being done at work, including the education and creative sectors, by accepting the wrong social contract with AI. And (perhaps most important) it debunks the philosophical prejudices that, since Descartes, have blocked us from fully valuing our own cognitive resources mas sentient animals. The argument here is not against AI. It makes no sense to be against AI (computational and calculative power) in the abstract. Clearly there are positive uses of such power – for example, to read X-rays and CT-scans for symptoms of life-killing diseases faster and at least as accurately as the available human doctors can. But it is equally absurd to be in favour of AI in the abstract, as most corporate hype and government policy are! What we need, urgently and everywhere, is to recalibrate the debate about AI in terms of what will achieve social goals, while avoiding social harms. To seek to change how societies define intelligence, expertise, and knowledge by management fiat runs deeply counter to all democratic principles. It is also likely to be against the interests of most embodied human intelligences in the long run. But it is not too late to negotiate a better social contract with AI – in every institution and every workplace. The urgency of doing so could not be clearer. --- ## [**DONATE TO PTP**](http://chuffed.org/donate/190851-pull-the-plug)