Why Big Tech wants more economists

PodcastThe Economics ShowAra KharazianOct 2

AI labs are building economic research programs and publishing data on AI use, productivity, employment, and wages, while retaining underlying datasets that outside researchers say they need to assess assumptions and workplace effects. Early findings include a 10% headcount increase among intensive AI adopters over two years, alongside evidence that faster research processes can leave scientists with more hypotheses than they can pursue. The discussion also raises questions about whether AI’s benefits will be unevenly distributed across countries and whether task-level gains will translate into broader economic growth or higher wages.

Artificial Intelligence Economic Growth Labor Market

Preview · 9 more points in the app

Read the Full Summary

Key Takeaways

Demand for evidence about AI’s economic effects is growing because those effects are not yet visible in productivity statistics, but AI labs retain the full usage datasets while outside researchers often receive sanitized or aggregated releases. The speakers argue that researchers need access to underlying data—not just published methods—to examine assumptions and omitted mechanisms.

AI can speed scientific tasks without necessarily increasing scientific output: scientists are heavy AI users, yet data checking and a backlog of promising hypotheses remain constraints on research. John Burn-Murdoch cautions that task-level productivity gains do not automatically translate into greater scientific value.

Evidence on AI’s workplace effects remains incomplete: one study found a 10% headcount increase over two years among intensive AI users, while John Burn-Murdoch says usage logs do not reveal workers’ experiences of possible work intensification, frustration, or job risk. Anthropic’s monitoring framework found no impact so far by mapping Claude’s work-related automation analysis to U.S. unemployment data, and is intended to identify concentrated disruption if it emerges.

TECHNOLOGY

Research and Trust — OpenAI has an economic research team and launched an exchange to collaborate with outside academics; Anthropic is doing similar work. The host links these research efforts to AI companies’ public-trust challenges and their efforts to make the case for their technology to governments and consumers.

“Beyond hiring teams of economists, the big AI labs are increasingly trying to fund and collaborate with researchers…”—

Influencer Marketing — AI companies and startups account for 28% of influencer-marketing spend, making them the second most likely sector to use it. The speaker suggests social media exposure can reach business buyers who encounter AI as consumers.

“I think it's less obvious that the second most likely sector to use influencer marketing is AI and software companies…”—

Scientists’ AI Use — A Google DeepMind study combined Gemini usage data with qualitative research, including interviews and surveys. It found that scientists are among the heaviest AI users, applying models to speed research and using specialized models for technical, domain-specific work.

“There was actually a piece of work by the team at Google, DeepMind, the other day, which I thought was really nice because what they did was they combined their…”—

ECONOMY

Demand for Data — The speaker says AI’s effects have not yet appeared in productivity statistics, creating demand for data on who uses AI and how firms use it. Governments, investors, financial markets, and the public want to understand the technology’s economic effects.

“You can imagine that AI is a very new and nascent technology. Everyone wants to know how it's going to affect the economy. It's not showing up in productivity statistics…”—

AI Firms’ Headcount — A study found that companies using AI intensively increased headcount by 10% over the two years after adoption; low-intensity users saw no statistically significant change. The guest said AI adopters were already early adopters, and the study compared them with firms that would adopt later. The groups had similar growth before adoption.

“And we found that companies that use AI intensely see a 10% increase in headcount over the two years following adoption…”—

Economists on Substack — John Burn-Murdoch says economists who have joined AI labs, including an economist at Google DeepMind, are publishing work or musings on Substack. That prompted him to look more widely for analysis.

“So it's interesting that you mentioned Substack because even some of the economists who have gone to work at their AI labs, I'm thinking, for example, of Alex who's gone…”—

AI Labs Hire Economists

The host frames the growing demand for economists at AI companies as a way to study AI's economic effects, while raising concerns about corporate influence over the research.

AI Adoption And Hiring

Ramp's data finds that intensive AI adopters increased headcount by 10% over two years, though the discussion cautions that fast-growing firms may be more likely to adopt AI.

AI Research And Workplace Effects

John Burn-Murdoch says AI-lab research varies in quality and argues that usage data needs to be paired with interviews and surveys to reveal how AI changes work.

Economists In Tech

The speakers compare AI labs' research programs with earlier tech hiring waves, noting that today's economists are studying broader questions about AI's effects on society.

AI Labs' Data Limits

The discussion warns that labs control and aggregate the most detailed usage data, leaving important questions about work intensity, frustration, and job displacement underexamined.

Anthropic Research Safeguards

Anthropic describes public methods and labor monitoring, while its economist warns that AI adoption is higher in wealthy countries and could widen global disparities.

Research And Public Narratives

John Burn-Murdoch sees promising economic research but wants more direct study of workers and says some lab publications are designed to shape public debate, not meet journal aims.