Daron Acemoglu, who won the Nobel Prize in 2024, is no heterodox economist. But now The Economist magazine has damned his work, because it challenges the untrammeled power of Big Tech elites.
Daron Acemoglu, winner in 2024 of the so-called Nobel Prize in economics, has just published a new book, What Happened to Liberal Democracy? Now The Economist — the paper, founded in 1843, that stands as the epitome of free trade and economic liberalism — has seized on the opportunity to settle a few scores. In a piece that has taken everyone by surprise, it leveled a scattering of theoretical criticisms at Acemoglu’s work and, above all, called into question the standing of a highly regarded economist. Perhaps the clearest evidence is that The Economist declares in its headline that Acemoglu is “oddly unconvincing,” and that his arguments about AI are pessimistic to the point of losing credibility. Small wonder that the target of the attack took it personally and felt obliged to respond in public.
Everything suggests that the ultimate cause of the falling-out is that Acemoglu’s academic work, which revolves around technological change and the role of artificial intelligence, offers a proposal to strengthen the working class. This is one of the points The Economist goes after, labeling his approach pessimistic and unspecific. The economist Mònica Martínez Bravo captured the tension well when she wrote of “what curious things happen when a Nobel laureate in economics writes a book arguing for putting the working class back at the center of economic policy.”
In what follows, I want to examine in what sense, and how far, Martínez Bravo is right. And along the way I want to head into the fascinating and worrying world of the socioeconomic consequences of artificial intelligence.
The Role of Automation
One of the great open questions in economics concerns the effects of automation, that is, of the replacement of human workers by machines. Does it destroy jobs? Does it depress wages? Does it alter the balance of forces between capital and labor? Two hundred years on, the debate has produced no consensus at all, only a range of approaches for thinking about a highly complex process. And to understand it properly, we have to go back to the early nineteenth century.
David Ricardo is regarded as one of the founding fathers of political economy. In the first two editions of his major work, the Principles, however, Ricardo had little in particular to say about automation, since he shared the conventional optimistic view of the economists of his day. It is important to grasp this way of thinking about automation because, in broad terms, it still dominates the profession. On this view, when machines replace workers in a production process, two effects follow. The first is the obvious one: a certain number of people lose their jobs. The second is that introducing the new machinery can create new jobs that absorb the previously displaced labor. In other words, for economists there is nothing here to worry about, because the economy regulates itself in such a way that the introduction of machines — the incorporation of new technology — is able to sustain the level of employment. And Ricardo thought more or less this, but only until 1821.
In the famous third edition of the Principles, Ricardo added a new chapter in which he judged the conventional view to be mistaken and acknowledged that “the substitution of machinery for human labour is often very injurious to the interests of the class of labourers.” The impact of this shift was enormous, and his disciple John Ramsay McCulloch complained to him bitterly on the grounds that it gave encouragement to the anti-machine workers’ movements. It was a time of enormous social unrest: the Industrial Revolution was unfolding across the United Kingdom and radically transforming the working class’s way of life.
In a 2024 academic paper, Daron Acemoglu and Simon Johnson addressed this historical moment precisely. In their account, Ricardo’s change of mind about machinery had to do with his election to the House of Commons. His seat allowed him to see firsthand what was actually happening in the cotton industry and so correct his earlier views, because what was happening bore no resemblance to the economists’ optimistic picture: the Industrial Revolution was severely degrading the living conditions of the working class. The paper, incidentally, leans heavily on the work of two giants of British Marxist historiography, E. P. Thompson and Eric Hobsbawm (though it also cites, by way of complement, the eminent mainstream historian Joel Mokyr). Acemoglu’s figures indicate that, with automation, the real wages of handloom weavers fell by at least 25 percent, with no possibility for those workers of moving into new, emerging sectors. Their living conditions were simply flattened by the introduction of machinery into industry. The paradox is that, because productivity was rising and wages were not, employers reaped enormous profits that pushed inequality upward. In a phrase which Acemoglu returns to again and again: prosperity was not shared.
The logic of this critical reading of automation had been built into Acemoglu’s work long before. It appears most clearly, in a form aimed at the general reader, in his 2023 book Power and Progress, cowritten with Simon Johnson, where they stress that introducing new technology is not automatically beneficial for society. Whether productivity gains are shared, they argue, depends on factors such as the type of technology, the institutions (norms, customs, laws), and the direction society chooses to give technological advances. Significantly, Acemoglu and Johnson accept that prosperity is shared only when those advances depart from elites’ self-interest. Hence their account of the improvements in working-class living standards during the second half of the nineteenth century, which they attribute to the democratization of the political system, the growth of trade unions and legislation protecting workers’ rights; all of them factors that forced technical change in a direction other than that of the elites’ narrow economic interests. Anyone familiar with the history of the labor movement will recognize a clearly subversive political element in this line of reasoning.
It would be a mistake to think that all this is merely economic history or nineteenth-century political economy. Acemoglu’s point of reference is always present-day technological change, and more specifically the adoption of artificial intelligence in today’s production processes. But he uses economic history to think about the contemporary process, asking (in Power and Progress) what happens if the mainstream economists are wrong: if AI fundamentally generates labor market distortions that widen inequalities in wages and employment; if its impact redistributes power toward elites; if it impoverishes billions of people in the developing world; if it entrenches ethnic or social prejudice; or if it destroys democratic institutions. His conclusion is that contemporary society needs what emerged in the nineteenth century: the growth of counterarguments and organizations capable of pushing back against conventional thinking. To give a fuller sense of where he stands, it is enough to note that one of his most recent papers, coauthored with David Autor and Simon Johnson, is entitled “Building Pro-Worker Artificial Intelligence.”
Not a Heterodox Economist
Although some have found him too much of a dissenter, it is worth remembering that Daron Acemoglu is no revolutionary in the classical sense, even within his own discipline. The first book of his to reach my shelves was not an essay but his 2009 textbook Introduction to Modern Economic Growth: a graduate text of formidable algebraic complexity that surveys the models of economic growth from a single vantage point, namely, neoclassical theory. This is the dominant theory in economics faculties, and other schools of thought have historically been ranged against it, such as the post-Keynesian and the Marxist, which are for that reason regarded as heterodox. Acemoglu belongs to the dominant school, and indeed he recently described debate about capitalism as “mindless.” In the worldview underlying his models, there are no relations of production; power enters as a category of bargaining strength and regulatory capture; and the view of institutions is thin and never structural. His concept of the working class is consequently not a Marxist one, but it leads to what he calls “working-class liberalism.”
Yet all this is exactly what makes him genuinely dangerous in the eyes of mainstream economists: Acemoglu reaches his “radical” conclusions from inside the profession and inside the mainstream. His entire academic output is a lucid and algebraically demanding reformulation of the neoclassical models of technical progress. And every step he has taken over the past twenty years, such as treating the division of income between capital and labor as endogenous to technical change, he has taken without drawing on the vast body of heterodox work that trod that path long before him. Acemoglu today holds a view of trade that would not seem strange to Marxist authors, and a view of income distribution that post-Keynesians such as Joan Robinson or Nicholas Kaldor would recognize. And we have already seen that his reading of certain aspects of nineteenth-century economic history has striking points of contact with Marxist historiography. The crucial point is that Acemoglu has arrived at conclusions very close to those of the post-Keynesian and even Marxist schools but through different concepts and methods. Mainstream economists cannot fault him for want of technical brilliance, which is amply established. But they are frightened by the political conclusions that follow from his deviation.
The maneuver had already been rehearsed with Joseph Stiglitz. The US economist received the Nobel Prize in 2001 for his contributions to neoclassical theory, in particular his analyses of asymmetric information, and during the 1990s he was known, among other things, for his critique of the hard version of ecological economics. All of this made him chief economist of the World Bank, one of the institutions charged with promoting the “Washington Consensus” among developing countries — that is, with imposing neoliberal policies. Before his term was up, Stiglitz left his post, and in 2002 he published Globalization and Its Discontents, in which he called much of neoliberal policy into question. It was a harsh diagnosis, and one that those of us who were then postgraduate students of international and development economics always found instructive: a neoclassical economist attacking the conclusions of his own old paradigm. As now, The Economist was the spearhead against what struck it as an inadmissible turn, remarking drily that Stiglitz could have written a good book about globalization, but perhaps next time, and conceding that he was a brilliant economist before proceeding to demolish his standing altogether. Nor was it the only example: Kenneth Rogoff, then director of research at the International Monetary Fund (the other neoliberal pillar of the international system), attacked Stiglitz fiercely and entrenched what became the settled view among mainstream economists that as a scholar Stiglitz was a genius with a wonderful mind, but as a politician he was far less impressive. It is the clearest precedent for The Economist’s article on Daron Acemoglu.
What we can expect over the coming years is a concerted effort, led by part of the mainstream community of which The Economist is merely the vanguard, to devalue Acemoglu’s reputation, not so much on technical grounds as on account of his political “eccentricities.” As we have seen, this is not the first case and probably will not be the last, but the underlying logic has been in place for a very long time. The aura of respectability within the mainstream economics community depends both on technical foundations (with a high degree of mathematical sophistication, which makes the work largely inaccessible to the general public) and on the “reasonableness” of one’s political conclusions. It is worth recalling that the Nobel Prize in economics is not really a Nobel Prize at all, since Alfred Nobel never envisaged such an award for economics. That such a prize exists today is the result of the interests of the Swedish central bank, which in 1968, amid a bitter confrontation with the Swedish Social Democrats (then pursuing their reformist road to socialism), needed to burnish the reputation of those who defended its own (conservative) political positions. The great Swedish economist Gunnar Myrdal received the prize in 1974; but his own reading of the episode was that, because he was a “radical” (his word), they made him share it with a “reactionary,” Friedrich Hayek. In 1977, Myrdal wrote a short article acknowledging that he had been wrong to accept it, that economics is not a science on a par with physics or the “hard” sciences and — take note — that the Nobel Prize in economics ought not to exist. What he was ultimately suggesting is that the award is heavily burdened with political values and principles.
What This Means for the Present
Daron Acemoglu’s work has, to my mind, many limitations. Perhaps the most obvious is that, for all its focus on technology and technical change, it entirely omits any consideration of energy. To discuss the transformation of the nineteenth-century British economy without mentioning the role of coal is odd. Equally problematic is the scant attention paid to the imperial dimension of British industrialization — that is, to the fact that colonial trade, slavery, resource extraction, and Britain’s position in the world economy were part of the historical conditions under which the Industrial Revolution took place. This connects with another point for which Acemoglu has been widely criticized, this time from the heterodox side, and which is present in earlier books such as Why Nations Fail: The Origins of Power, Prosperity, and Poverty, cowritten with James Robinson. Acemoglu’s account of economic development is institutionalist, treating factors such as innovation, democracy, and types of government as the central variables determining which countries grow and which do not.
But his interpretation of artificial intelligence, his emphasis on the task model — it is not jobs that are automated but tasks — and his insistence that the effects of automation depend on power and on institutions such as the law, are correct and crucial for our present moment. It is true that other traditions can offer sharper and more structural accounts of the role of technology, but Acemoglu’s originality and brilliance lie in having reached the same destination as heterodox authors, albeit by another route. Complementarities between the different traditions are well worth exploiting. His insistence across several papers that technology can be (mis)used to tighten control over workers (something that AI is already doing) could gain a great deal from the entire Marxist tradition that begins with Harry Braverman’s superb book, Labor and Monopoly Capital, and revolutionized labor market theory. The same goes for the role of income distribution, which is not only a moral question but also one of economic efficiency (particularly in models where demand plays an important part).
Finally, many unresolved aspects of Acemoglu’s work will require development over the coming months and years, and they are decisive for public policy. His central proposition is that technology must be directed if it is to promote shared prosperity. His critics, including The Economist, point out how vague this is in practice. On this they are right: Acemoglu and his coauthors sketch a course that involves intervening in the outcomes of the self-regulating market in order to steer technology “somewhere else.” The historical comparison implies that the effect sought is the strengthening of trade unions, democracy, and counterpower against elites (read: the tech oligarchs), but the “how” of achieving all this is still missing.
Acemoglu also insists that society has alternatives to deploying technology in ways that will seriously hurt the working class. He accepts that employers seek to raise their profits and increase average labor productivity, even where that deployment produces no social benefit (in his terms, where it does not raise the marginal productivity of labor). A structural analysis will object that an employer adopts a given profit-maximizing technology because competition compels him to, so it is unclear how things could be otherwise without recourse to classical instruments such as state regulation. At root, the ordinary grammar of any process of “directing technology” is that of planning, the state, and social protection; and that is probably where the next round of arguments in the field will be found.
In sum, I hope to have shown why I believe, along with other economists, that The Economist’s attack on a widely respected and brilliant economist can be explained. Acemoglu’s problem, from the mainstream’s point of view, is that he has used mainstream economics’ own conceptual tools to show that technology and the market do not automatically produce shared prosperity. To accept that argument is to call into question every promise made by the promoters of AI (and also its associated costs, in which energy once again plays a central part), and to argue for public policies utterly unlike those defended from the centers of power tied to the big technology firms. To my mind, that is a highly fruitful path, and it is clear that we need more shared prosperity, which is to say, more of the sort of interventions and policies that Acemoglu proposes.