Are economists making themselves too useful in the AI boom?

Economists are a fashion-conscious bunch. In the 2000s and 2010s it was cool to work on and with new digital companies. In 2020 they swung towards Covid-19 research with the enthusiasm of an Instagrammer learning to bake bread. And now, covering AI and working with the AI labs is hot. Although some of the risks have echoes of the past, this time feels different.

One difference is the vast scope of questions AI is raising. It’s one thing to tinker with taxi or hotel markets, and quite another to explore a technology that could fundamentally reshape society. In the 2000s and 2010s, the digital revolution attracted star researchers and raised new questions. But because the AI revolution is bigger and broader, today there is a “completely different scale of sucking sound”, says Ben Golub, an academic with an AI start-up of his own.

As AI presents economists with both a tool of analysis and a topic of study, one interesting trend has been elite economists either collaborating with AI companies, or working for them directly. This year there has been a “discontinuous change in perception”, as AI labs are now seen as “super-high value places to be intellectually”, says Golub. It helps that they have detailed data describing AI usage, which is key to estimating its economic effects.

The broader hunt for economists by AI-related start-ups could be cynically dismissed as an elaborate PR ploy. Credible analysis of proprietary data from highly educated nerds is more likely to get media attention than a fluffy press release. After sharing analysis of business spending on AI tools, Ara Kharazian, Ramp’s chief economist, reported being inundated with requests for advice from AI start-ups looking to start their own economics teams.

But the bigger AI labs seem to have grander ambitions, to inject evidence into conversations about their technology. Aware of AI’s seismic potential, researchers at Google, Anthropic and OpenAI have been tasked with informing the debate over AI’s effects on labour markets, global inequality and the macroeconomy. One Google researcher framed the research agenda to me as being a part of corporate responsibility.

The obvious risk is that closer collaborations will distort the direction of economic research. Why would the AI labs want to release research showing that their products make work less meaningful or more intense? Would they be as keen to support researchers warning that AI will discourage investment in skills as they would those who seem optimistic about its effect on entrepreneurship?

Potential “capture” was a challenge during the last tech boom, as tech companies controlled who could access their vast troves of data, and had private interests in how it was used. One recent study noted that European papers published using platform-provided data were disproportionately about the effects on consumers, rather than, say, labour rights. (Admittedly, they only picked up 31 such studies, which were mostly about Uber.)

Quizzing some economists who are part of this latest tech wave, I heard cautious optimism that conflicts of interest could be manageable. Although it’s true that the labs have probably drawn economists who are likely to think that AI is a big deal, those researchers don’t agree among themselves how it will play out. One pointed out that Dario Amodei, Anthropic’s CEO, hasn’t exactly been promising that AI would deliver rainbows and sunbeams.

Within the AI labs, economists have been thinking about how to ensure their own credibility. Earlier this year Anthropic’s economics team published a framework for measuring AI-fuelled labour market disruption. Chief economist Peter McCrory explained that by tying their hands to a particular method, it shouldn’t be possible to suppress bad results. The labs also seem keener to share data publicly than some of their predecessors, though not enough to erode the insiders’ advantage.

As for the collaborations, it’s not yet clear how onerous the research approval processes will be. One researcher said that he didn’t think that OpenAI’s funding grants were large enough to get big cheeses to suppress or change results. Economists have reputations to maintain; several of those working with the labs are only on leave from their university positions.

All that said, as the public backlash against AI grows, so too will pressure on the AI labs to shape the narrative around their technology positively. And where those pressures are strongest, the presence of researchers with distance from the labs will be vital. After all, there’s one final difference between this tech boom and the last: the stakes are higher this time.

[email protected], @SoumayaKeynes

The Economics Show with Soumaya Keynes is a podcast from the FT bringing listeners a deeper understanding of the most complex global economic issues in easy-to-digest weekly episodes. Listen to new episodes every Friday on Apple, Spotify, Pocket Casts or wherever you get your podcasts