Former CBO chief Doug Elmendorf: AI job losses are 'very likely' over the next two decades, with one scenario putting 3 million out of work at a time

Doug Elmendorf, the Harvard Kennedy School economist who ran the Congressional Budget Office, says the calm in today's job market tells us little about AI job losses to come. "What we've seen so far in the labor market from artificial intelligence has very little predictive power for what we're going to see in the labor market because of AI in five years, or 10 years, or 15 years," he told the Harvard Gazette, in a piece published on 29 September 2026. With Karen Dynan of the Kennedy School and Louise Sheiner of the Brookings Institution, he wrote a National Bureau of Economic Research working paper this spring setting out four scenarios, from a moderate boost to GDP with little loss of workers to much faster growth with persistently high unemployment. The authors do not rank them. In the scenario they compare with the China shock, which other researchers put at 1.5 to 2 million jobs between 2000 and 2007, about 3 million people, roughly 2 percent of the labour force, would be out of work at any given time. "I think it's very likely that there will be significant job losses from artificial intelligence in the coming couple of decades, but that most, not all, but most, of those people will be able to find new work," Elmendorf said. He does not expect appeals to employers to hold back the cuts: "Individual companies are going to be focused on what's good for their company... And for a number of companies, that will mean the use of AI instead of workers for some things." The paper's answers are for government: changes to taxes, the safety net and unemployment insurance, subsidised private employment, shorter working weeks, more retraining and wage insurance for laid-off workers who take lower-paid jobs. In the same piece, Harvard Business School's Joseph Fuller says 41 percent of work tasks can already be automated or augmented by AI but only about a third of firms' AI experiments succeed, and his colleague Raffaella Sadun calls automating away investment in staff "a terrible trap." Elmendorf frames his case as insurance: "Policymakers should be prepared for much bigger economic changes than we've seen so far from AI."