When Project Syndicate published Reclaiming Democracy From the Market in early March 2026, the headline read like a co-authored manifesto. It was not. It was an interview. Daron Acemoglu — the MIT economist who shared the 2024 Nobel Memorial Prize in Economic Sciences and co-wrote Power and Progress with Simon Johnson — was the one asking the questions. Michael Sandel — the Harvard political philosopher behind The Tyranny of Merit — was the one answering.1

I want to spend some time on that asymmetry. The form of the publication is doing more work than its content, and most of the readings I have seen miss it.

What was actually argued

The substantive prescription is unusually concrete for a piece in this register. Three institutional moves recur. The first is what Acemoglu and his co-authors have elsewhere proposed as a more symmetric tax structure on labour and capital — equalising effective tax burdens so that the code stops quietly subsidising the substitution of machines for people.2 Effective US tax rates on labour have run well above 28 per cent for years, while accelerated-depreciation provisions have pushed the effective rate on capital invested in equipment and software far lower; that gap, summed over decades, produces a structural incentive to automate that has nothing to do with whether automation is a good idea in any given case. The second is technological co-determination: an extension to AI of the German Mitbestimmung tradition, in which workers elect representatives to the supervisory boards of large firms.3 The third is sectoral bargaining — collective negotiation across an industry rather than firm by firm, which is closer to how labour relations are organised in much of Northern Europe than to the post-Wagner-Act American model.

What both interviewer and interviewee push back against, in unusually direct language, is universal basic income. Sandel objects on the grounds he has set out at length elsewhere: a UBI treats people as passive recipients and abandons the normative weight he places on the dignity of work. Acemoglu objects that re-distributing the gleanings of automation is the wrong frame in the first place — the real question is which technologies get built and what they substitute for, not how to compensate the people they displace afterwards.

The framing they offer for the bundle is the move from distributive justice — re-allocating after the market has produced its outcome — to a pre-distributive posture: shaping markets and technologies before the outcomes harden.

Why the interview was an interview

Here is what I think is going on with the form. Read carefully, the policy content is almost entirely Acemoglu’s. Symmetric taxation, co-determination, sectoral bargaining — these are positions he has developed at length in his own technical and popular work. Power and Progress spends large stretches on each of them. So the interesting question is: why does an economist of Acemoglu’s standing need to interview a moral philosopher to advance a policy programme he has already, on his own, defended in print?

The answer, I think, is that economics alone cannot beat UBI on its own terms. Within a strictly welfarist or efficiency-maximising frame, UBI tends to win comparisons with elaborate institutional reforms: it is administratively cheaper, more legible, and harder to capture. The economist’s vocabulary makes the comparison roughly: which intervention raises measured welfare per dollar at the lowest deadweight loss? On that yardstick, building out co-determination structures, retraining sectoral bargaining institutions, and rewriting depreciation schedules is a long, politically expensive march. Sending people money is faster.

What Sandel’s “dignity of work” does is change the yardstick. Once the comparison is no longer “which option maximises welfare per dollar” but “which option preserves the conditions under which a person can hold a position of recognised social contribution,” UBI starts to look like something that fails on its own merits, not just on cost-benefit grounds. The new yardstick is not metrically commensurable with the old one. It refuses, in a way that economics on its own struggles to refuse, to treat work as merely instrumental to consumption.

So the interview architecture is doing a specific structural job. It allows the economic positions to keep their concreteness while borrowing a normative anchor from outside economics. The philosopher is not contributing the policy; the philosopher is contributing the measure by which the policy is to be judged. One might call it a borrowed yardstick.

Whether one finds this welcome or worrying depends on whether one trusts the borrowing direction. In one reading, this is a healthy interdisciplinary correction — economics has been operating with too thin a normative vocabulary for decades, and philosophy is offering a richer one. In another reading, it is a sign of how exposed even Nobel-laureate economics now is to UBI’s rhetorical pull: the discipline that was supposed to be the most rigorous arbiter of policy choice has begun to outsource its tie-breakers.

The substrate problem

There is a more uncomfortable question that the interview does not really address. The institutional reforms on offer — co-determination boards, sectoral bargaining, expanded apprenticeship — all assume that an organised body of workers exists to occupy the seats they create. Co-determination only does its work if there are workers organised at the firm and sector level. Sectoral bargaining only stabilises wages if there are unions with members.

Some of the labour-market data that has emerged over the past year suggests that the substrate of organised labour is not steady. Entry-level white-collar work — the layer through which young people have historically passed on their way into stable, organisable employment — is the layer where AI displacement appears earliest and most visibly. Yale’s Chief Executive Leadership Institute reported in early 2026 that the most consistent pattern across the data they track is hiring slowdown rather than mass layoffs, concentrated on first jobs.4 An Anthropic study released around the same time, which combined a measure of LLM capability with real-world usage, found suggestive evidence that hiring of younger workers had slowed in occupations most exposed to AI tooling.5 If the layer thins out durably, then the institutions Sandel and Acemoglu are reaching for could end up arriving at a workplace whose organised workforce has already partially evaporated. The seats are built; nobody is left to sit in them.

This is not a fatal objection — institutions can sometimes reconstitute the constituencies that depend on them — but it is the gap I see most clearly in the prescription as offered. Power and Progress-style institutionalism tacitly assumes the twentieth-century industrial substrate of workers organisable at scale. The shape of the displacement underway is partly attacking that substrate itself.

The risk of a new meritocracy

The other gap is internal to Sandel’s own framework. The Tyranny of Merit targets credentialism — the social sorting that runs through degrees, test scores, and elite-school admissions.6 But suppose Acemoglu’s “human-complementary AI” succeeds and we end up in a workplace where the most productive workers are those who can collaborate effectively with AI tools. A new axis of competence emerges almost immediately: workers who can effectively co-pilot with AI versus those who cannot. That distinction will sort people into compensation tiers as ruthlessly as the SAT once did.

Sandel’s “dignity of work” is conceptually strong, but it is not obviously protected against the implementation question of which work gets dignified. If, in practice, the dignified worker is the AI-fluent worker, and the AI-fluent worker disproportionately comes from the same demographics that already won the credential lottery, then the new vocabulary will be reproducing the old sorting under a different name.

What to watch

The question worth asking, I think, is whether the borrowed-yardstick move is repeated. If it is — if economists of Acemoglu’s calibre keep needing to invite philosophers in to do legitimation work that economic vocabulary cannot do for itself — that is a structural fact about contemporary policy argument worth tracking on its own terms. It would tell us something about which discipline is currently the load-bearing one for the AI-and-labour debate, and which is performing the older role of background justification.

Equally worth watching is whether the institutional prescriptions in the interview travel — whether anyone seriously attempts, in the United States in particular, to build out the German-style co-determination scaffolding for AI that the interview gestures at. The political-economy obstacles are formidable, and the interview itself is silent on who exactly is supposed to do the building.

For now, the most useful thing the interview does is expose its own architecture. An economist asks. A philosopher answers. The questions encode the policy. The answers encode the warrant. The reader is asked to take both as a single position. Whether that combined position holds, or whether it is the surface of a deeper instability inside the economics-of-AI debate, is the question I find I cannot yet settle.

  1. Acemoglu, D. & Sandel, M. “Reclaiming Democracy from the Market”. Project Syndicate, March 2026. Accessed 2026-05-09. ↩

  2. Acemoglu, D., Manera, A. & Restrepo, P. “Does the U.S. Tax Code Favor Automation?” Brookings Papers on Economic Activity, Spring 2020. See also Acemoglu, D. & Johnson, S. Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity. PublicAffairs, 2023. Accessed 2026-05-09. ↩

  3. Federal Ministry of Labour and Social Affairs (BMAS), “Co-Determination”. 2019. See also Wikipedia, “Codetermination in Germany”. Accessed 2026-05-09. ↩

  4. Sonnenfeld, J. & Tian, S. “The Real Job Destruction from AI Is Hitting Before Careers Can Start”. Yale Insights, 2026. Accessed 2026-05-09. ↩

  5. Anthropic, “Labor market impacts of AI: A new measure and early evidence”. 2026. Accessed 2026-05-09. ↩

  6. Sandel, M. The Tyranny of Merit: What’s Become of the Common Good? Farrar, Straus and Giroux, September 2020. ↩