Harvard Researchers Say AI's Job Impact Is Still Ahead, and Companies Are Slow to Adapt

Harvard researchers argue that AI's effect on employment has not yet arrived and may take years to appear. Doug Elmendorf of the Kennedy School, with Karen Dynan and Louise Sheiner, wrote an NBER working paper sketching four scenarios for AI's economic impact. Elmendorf says the labor market so far has little predictive power for conditions five to fifteen years out, and he expects significant job losses over coming decades, though he thinks most affected workers will find new work. Joseph Fuller of Harvard Business School, working with Accenture Research, estimates that 41% of work tasks can be automated or augmented by AI. He says only about a third of firms' AI experiments succeed, partly because companies do not train employees formally. Raffaella Sadun of the same school calls cutting human capital to save money a "terrible trap." Fuller describes redesigning entry-level roles so junior hires keep learning. Elmendorf argues that policy, not company goodwill, must handle the disruption. It is worth reading because it separates the dire forecasts from the evidence and lays out the trade-offs executives face.