Federal Reserve Chair Kevin Warsh opened an AI productivity and jobs conference in San Francisco on November 16, 2026, bringing academic researchers and business leaders into the central bank’s assessment of technological change. For universities studying AI’s economic effects, the event connects questions about faster scientific idea generation with evidence on adoption, employment and economy-wide productivity.
Warsh sets out three possible AI outcomes
In his published opening remarks, Warsh said Federal Reserve staff had begun using stylized, model-based frameworks to examine different contingencies. He described the scenarios as illustrative and explicitly distinguished them from forecasts.
One scenario combines stronger productivity with gains shared by workers and capital owners, sustained growth and employment close to its full-employment level. A second combines faster output growth with displacement and labour-market frictions. In the third, disappointing returns after substantial investment contribute to a slowdown and higher unemployment, without the anticipated productivity gains.
Reuters’ report by Ann Saphir and Howard Schneider corroborates the range of outcomes discussed. Associated Press coverage also identifies the remarks as the opening of the one-day conference. The development is a public evidence-gathering discussion about AI’s effects, with the eventual economic outcome left conditional in Warsh’s remarks.
Stanford and MIT researchers appear on the programme
The San Francisco Fed’s official agenda names Stanford economist Nick Bloom, MIT computing dean Dan Huttenlocher, Stanford president Jonathan Levin and Brookings president Cecilia Rouse for a panel on interpreting the economic signals of AI adoption.
The programme also includes a discussion of adoption at the technological frontier, with leaders from Palo Alto Networks, Stripe and Affirm. A separate implementation panel brings together Indeed, Ambi Robotics, Pearson Packaging Systems and Providence. The sequence connects the technology’s capabilities with evidence from organisations putting it into practice.
The organisers are the Federal Reserve Board and the Federal Reserve Bank of San Francisco. Their event page says the conference informs a task force assessing new general-purpose technologies. It places that work within five task forces announced in June, supported by Fed staff and operating independently to provide rigorous findings and candid feedback to the Federal Open Market Committee.
Scientific idea generation is part of the growth question
Warsh’s argument reaches beyond automation of existing production. He suggests AI could change how ideas are generated and the speed of innovation. That proposition makes the productivity of research itself relevant to the central bank’s inquiry, alongside the employment consequences of adopting new tools.
His speech cites Are Ideas Getting Harder to Find?, a 2020 American Economic Review paper by Nicholas Bloom, Charles Jones, John Van Reenen and Michael Webb. The authors distinguish the effective number of researchers from research productivity. Their historical evidence finds rising research effort alongside declining productivity across a range of settings.
The paper examines semiconductor progress, agricultural yields, medical advances and firm-level data. Those cases establish the background to Warsh’s question about whether AI could change the production of ideas. They also make the measurement problem concrete: more research activity and more useful output are separate quantities that an evaluation must connect.
CASRAI’s earlier coverage of the Fed’s AI evidence working paper concerns a separate publication. The November 16 conference adds a new forum in which academic and industry perspectives inform the task force’s work.
How the stakeholders frame the inquiry
Warsh said the Fed’s responsibility is to prepare for technological change and its implications for policy. Reuters frames his remarks around the range between broadly shared gains and costly disruption, while AP highlights both innovation’s potential and employment risks. These accounts describe conditional possibilities rather than a measured result from the conference.
What research offices can take from the discussion
The following are editorial implications for university evaluations. Research offices can distinguish tool adoption, time spent on tasks, validated research outputs and changes in staffing. Keeping those measures separate makes an institutional assessment more informative than treating usage alone as evidence of improved research productivity.
Economists studying total factor productivity can likewise identify the mechanism a study measures: faster completion of a task, greater output from the same inputs, or wider diffusion of an innovation. Each addresses a different part of the growth question raised at the conference.
For sponsored projects, a reproducible AI experiment can record researcher effort, model configuration and validation costs alongside results. That gives research managers evidence for decisions about tools and support, while preserving the distinction between a local project outcome and a claim about national economic growth.
What happens next
The organisers describe the conference as input to the task force’s work for the Federal Open Market Committee. Its stated mandate is to follow the evidence and produce rigorous findings. For researchers following that work, the next substantive development will be the evidence and analysis the task force makes available, rather than an assumption that one of the illustrative scenarios has already occurred.
Sources
- Federal Reserve Board: Kevin Warsh, Opening Remarks, November 16, 2026; checked 2026-11-17.
- Federal Reserve Bank of San Francisco: Productivity and Jobs Task Force Conference, November 16, 2026; checked 2026-11-17.
- Reuters: Warsh says emerging technologies will have significant impact, with risks, November 16, 2026; checked 2026-11-17.
- Associated Press: Warsh discusses AI’s potential economic impact, November 16, 2026; checked 2026-11-17.
- Bloom, Jones, Van Reenen and Webb: Are Ideas Getting Harder to Find?, American Economic Review, April 2020; checked 2026-11-17.







