AI in action · Regulation
Fed's Jefferson: AI Is Still Faint in Productivity Data
Fed Vice Chair Philip Jefferson said on November 13, 2026 that AI's contribution to US productivity is still limited, though it could grow substantially.
In 1987 the economist Robert Solow quipped that "you can see the computer age everywhere but in the productivity statistics." Federal Reserve Vice Chair Philip Jefferson opened his speech at the University of California, San Diego on November 13 with that line, and argued it fits AI today.
His evidence is a mix of fast adoption and slow results. Surveys show 40% to 50% of working adults use generative AI at work, but only a modest share use it daily. About 25% of US firms used AI in at least one business function by mid-2026, close to 40% in knowledge-heavy sectors such as professional services and information and around 10% in mining and in leisure and hospitality. At the firm level, one survey he cited put the gain at roughly 0.5% in sales per employee over three years.
US labour productivity has picked up: growth averaged 2.2% a year from 2020 to 2025, against 1.5% in the previous business cycle. Jefferson was careful not to credit AI for it, noting that the improvement started before chatbots became popular in 2022. For comparison, information technology investment added about three-quarters of a percentage point a year to productivity growth in the late 1990s. His summary of AI so far was "a contribution to aggregate productivity that is limited, but with the potential to grow quite substantially."
For markets, the policy passages mattered more. Jefferson said the AI build-out is adding to price pressure now, through demand for items such as memory chips, at a time when headline PCE inflation was 3.6% in September and core PCE 3.1%. He described the trajectory of inflation as "the more important signal" for the Fed's next decisions. If AI does lift productivity permanently, he added, the neutral interest rate could rise, which would make the current stance less restrictive than it looks.
He also drew a limit around what the Fed can do. If AI displaces workers faster than they can retrain, monetary policy can soften the cyclical part of the job losses, but "it cannot address the structural factors that determine long-run rates of employment."
Sources: Federal Reserve — Jefferson, New Technology, Old Questions: Productivity Puzzles and the Macroeconomy (Nov 13, 2026); Federal Reserve — 2026 speeches