Nobel Laureate Daron Acemoglu on ‘brainless’ AI discourse, fantasy of capitalism and Gen Z revolution

And when asked how much of the current AI discourse he finds intellectually serious, he doesn’t hesitate: about 20%.

“I find all of this discussion of capitalism so brainless,” Acemoglu told Fortune, insisting that we should be focused on the “enormous increase” in corporate power and monopoly instead. “That’s what we should be talking about. What we should be talking about is the displacement and unequalizing roles of AI.” When asked how much of the discourse he finds, in his words “brainless,” he barely paused. “About 80%,” he said. He clarified that the thinking is rather speculative or close to fictional, not stupid per se.

“Unfortunately, a lot of the left is a big contributor to that,” he added, stressing that it’s a central point of his forthcoming book What Happened to Liberal Democracy? “The success of liberal democracy was rooted in social democratic, center-left ideas, and governments playing a leading role. And that space cannot be filled by stupid ideas and by being completely unaware of, you know, what AI is doing, what are its capabilities, what are its implications, nor could it be filled by, Frankfurt School-influenced quasi-Marxist oppressed/oppressor dynamics applied to everything.”

He added acidly that he’s grown sick of the phrase “colonizing AI” as an example of unhelpful Marxist rhetoric, detached from a center-left lens that would actually be practical and helpful. It’s vintage Acemoglu, extending his decades-long argument that the health of economies and the health of democratic institutions are inseparable — and that AI is now stress-testing both simultaneously.

‘Capitalism is a completely useless word’

Ask Acemoglu where he stands on capitalism and he’ll redirect the question entirely.

“I don’t like the term capitalism,” he said. “It makes it sound like there is a uniform model that includes Sweden, Egypt, Argentina, Honduras, the United States, South Korea, Japan. There’s no overlap between these economies, how they are organized.” The only overlap he sees is that they have markets, “but so did the Soviet Union.”

His preferred frame, developed across Why Nations Fail and The Narrow Corridor with co-author James Robinson, is inclusive versus extractive institutions. The question isn’t whether a country has markets, but whether its economic and political rules broaden participation and reward innovation — or whether they concentrate power at the top and extract value from everyone else.

Seen through that lens, AI is not troublesome in its own right, but rather whether it is positioned as inclusive or extractive. Today’s AI hyperscalers, he argues, fit the extractive mold almost perfectly: concentrated ownership, regulatory capture, and a business model that extracts data and attention at scale.

Instead of reckoning with that, he said, we get all kinds of other talk about whether capitalism is mutating into technofeudalism, or whether AI will automate away every job in existence. “People are saying such stupid things. I can’t believe it.”

The productivity illusion

Acemoglu’s skepticism about AI’s economic upside isn’t contrarianism — it’s grounded in a framework he’s applied to every major wave of automation for decades.

Productivity gains from automation, he explained, only materialize if machines can do tasks significantly cheaper or better than humans. If the improvement is marginal, or if integration costs eat into gains, the math doesn’t add up — even if the automation is widespread. “It’s not that you cannot get big productivity gains from automation,” he said. “It is that it’s not as easy as sometimes it’s presumed.”

What would actually move the needle? True “human complementarity,” Acemoglu insisted, would be AI that enables workers to do things they simply couldn’t do before, expanding the range of tasks and services on offer, rather than just accelerating existing ones. He turned media critic briefly: “Podcasts massively expanded the demand for news,” he noted. If AI can do what he calls “new tasks” — versions that were not previously available — “that is the real pathway to true human complementarity, not just enabling you to do what you were doing before in a better way, or in a faster way.”

Acemoglu nodded when Fortune mentioned his finding that most research on AI productivity is overblown because it overwhelmingly focuses on easy, well-defined tasks where context is clear. These are not representative of the economy, which simply isn’t set up with so many of those, and AI is just not great for hard tasks yet. “You need new tools, sort of a tool that reliably understands and distills the best research, and is not swayed by the worst research in a particular field, and provides that to you in a context relevant and an accurate manner, and allows you to interrogate it.”

The sharpest version of his argument cuts even deeper: the productivity gains that AI bulls are penciling in don’t just require better models — they effectively require artificial general intelligence. For genuinely huge productivity gains from automation, “then we really, really need something close to AGI for that,” Acemoglu said, referring to the concept of artificial general intelligence. “So that’s why AGI is not just a theoretical issue — it’s really relevant for these productivity projections.” He’s skeptical we’re close. Current models, he argues, perform badly across too many dimensions of real-world work — they can’t read a room, they can’t connect non-obvious dots across domains, and they fail precisely where human judgment is most valuable. The gap between what LLMs do well in demos and what they do reliably in complex, high-stakes professional environments remains, in his view, far wider than the industry’s marketing suggests.

The revolution risk

Acemoglu added that the Fortune 500 should hope that he’s right, paradoxically, that AI won’t be that useful.

“If it were the case that 30%, 40% of new university graduates can’t find jobs,” he said, “what would that do to democracy and social peace? Wherever that has happened in the past, you’ve had revolutions.”

Revolutions, he added quickly, are inherently unpredictable, shaped by the interplay of repression, redistribution, and the ambient attitudes of a generation. Social media adds a new variable that history offers no reliable guide to. “In the past, youth did not have Instagram, TikTok, and Twitter,” he said. “Perhaps that changes things. I have no idea.”

But the direction of concern is clear. A generation of workers who trained and credentialed for an economy that AI has since restructured — and who feel economically stranded — is a constituency that has historically not stayed quiet. The grumbling at this spring’s commencement ceremonies, he suggested, may be an early signal.

What would fix it

Acemoglu’s critique comes with a prescription, though he’s frank about how far the current moment is from acting on it.

The U.S. needs to have a genuine conversation about what is socially desirable from AI — not just what is technically possible or financially profitable for a handful of hyperscalers. That conversation, he argues, has to center on wages, jobs, shared prosperity, and “meaningful, dignified lives for workers.” It also has to include serious global governance — including cooperation with China, which he says is ahead of the U.S. in integrating AI into manufacturing, robotics, and commerce, even as it lags on large language models.

“I think that [U.S.-China collaboration] would be so beneficial,” he said. “We need global governance for AI. We also need the ‘AI race’ not to get out of control. And we need the two sides to share best practices on things that are useful for humanity,” he said, mentioning disease control, productivity, shared best practices and global safety regulations. The current geopolitical climate, he acknowledged, makes that nearly impossible. “The only bipartisan issue in the United States right now is China bashing,” he said, adding that it was that way during the Biden era.

The intellectual failure, in his view, runs deeper than policy. It’s a failure of imagination — an inability to articulate what a genuinely human-centered AI future would look like, and the political will to demand it.

“We’re all so blindly taken in by what OpenAI, Anthropic, and a few other hyperscalers are offering,” he said, “because we haven’t articulated a reasonable alternative.” Squint and you hear the old phrase from the Paul Newman classic Cool Hand Luke: what we have here, gentleman, is a failure of imagination.