About three weeks ago, The Economist published an unsigned editorial sharply criticizing Nobel Prize-winning economist Daron Acemoglu, setting off a debate in the economics world. The controversy was fueled in part by the stature of the two sides: a scholar widely regarded as one of the world’s most influential economists pitted against one of its oldest and most prominent business and current-affairs publications.
The episode has brought renewed attention to criticisms of Acemoglu’s research. And while The Economist itself made an unconvincing case for those criticisms, many economists believe there is considerable substance to at least some of them.
The article argues that Acemoglu’s research, and the conclusions he draws from it, are “unconvincing.” In recent years, much of Acemoglu’s academic work and public advocacy has focused on the development of artificial intelligence and its implications for the distribution of wealth. It is clear that The Economist is intent on challenging his authority on the subject.
Yet rather than directly confronting Acemoglu’s arguments and explaining why they are wrong, the magazine strings together a series of criticisms, some substantive, others less so, in an effort to undermine his credibility more broadly.
AI That Works for Workers
Acemoglu’s argument is that, much as with previous technological revolutions, there is a risk that a large share of the gains from artificial intelligence will accrue to a narrow class of capital owners, while many workers suffer from automation and declining bargaining power.
Across a series of papers, he has argued that public policy should be used to steer the development and deployment of new technologies in ways that benefit workers as well. Without such a shift, he warns, AI could inflict significant social damage and even undermine democracy.
In Acemoglu’s view, the concentration of wealth will ultimately lead to a concentration of political power in the same few hands. People who earn their living primarily through work rather than through ownership of capital will see their political influence diminish.
At the same time, Acemoglu argues, artificial intelligence could dramatically expand the ability of governments and corporations to shape the public sphere. AI can make it cheaper and easier to create and disseminate misleading information, while also enabling sophisticated systems for monitoring citizens — developments that could erode the conditions necessary for healthy and fair democratic debate.
Acemoglu’s proposed solution is to shape the development of AI so that it expands human capabilities and creates new, productive tasks for workers. That could mean, for example, developing systems that enable a nurse, electrician or technician to perform more sophisticated work than they can today.
For Acemoglu and his co-authors, the direction of technological development is not predetermined. Public policy can influence it through research funding, tax incentives, regulation and protections for workers’ bargaining power.
It is precisely this idea — deliberately shaping the direction of technological development — that The Economist opposes. Its broadside against Acemoglu appeared alongside a review of his new book, which the magazine chose to frame largely around the question of artificial intelligence, even though Acemoglu says only about half a chapter of the book is devoted to the subject.
Personal Attacks Without an Argument — and Substantive Criticism on Other Matters
The Economist attacks Acemoglu in unusually harsh terms. Yet his position on AI is never presented in anything approaching meaningful detail, much less addressed head-on.
Instead, the article quickly arrives at the claim that, after “a few drinks,” some economists will reveal what they really think of Acemoglu. It then quotes an anonymous economist criticizing both his work and the policies he advocates. The passage itself drew criticism, both because the economist was not identified and because its tone was widely regarded as unserious.
The article also raises more familiar and substantive criticisms of Acemoglu’s scholarship. But these have little to do with his arguments about artificial intelligence — the very arguments The Economist ostensibly seeks to challenge.
One example is a well-known academic critique by economist David Albouy of Acemoglu’s influential 2001 research, which played a crucial role in the work for which he was awarded the Nobel Prize. That research argues that the quality of economic institutions helps explain a substantial share of the differences in economic outcomes among countries. Albouy argued that the data used by the researchers had been constructed in a problematic way and that correcting those problems effectively eliminated the result.
It is an important criticism. But, again, it has nothing to do with Acemoglu’s arguments about artificial intelligence.
In his response to the article, which The Economist published in full, Acemoglu argued that the magazine’s critique ultimately demonstrates the weakness of its own case. Out of a body of work spanning 33 years and hundreds of papers, he noted, the magazine points to several familiar intellectual disputes over institutional theory and a broad quotation from a blogger who has criticized him for years. In another instance, it presents as a weakness a caveat that Acemoglu and his co-authors themselves explicitly acknowledged in the paper’s conclusion.
Disputes of this kind, Acemoglu argued, are a normal part of academic research and do not justify the broader insinuation that his scholarship is unreliable. If this is the strongest case that can be assembled against a body of work that extensive, he suggested, then it is The Economist’s critique that is “unconvincing.”
Acemoglu added that academic criticism is essential and that he welcomes it. Curiously, however, his response did not address Albouy’s criticism, despite its inclusion in The Economist article. Albouy, for his part, has argued that Acemoglu has never shown a willingness to acknowledge the flaws identified in his work.
Rather than focus on Acemoglu’s arguments about artificial intelligence and explain why it opposes his proposals, The Economist devotes much of its article to an effort to undermine his standing more generally.
The result is strikingly clumsy, particularly for a publication with The Economist’s long-established reputation. More than anything, the article reads as an expression of deep-seated, almost instinctive opposition, but one that is barely argued, to the idea that public policy should play a role in steering technological development so that its benefits are shared with workers.