AI is not killing jobs, US study finds (www.ft.com)

A new U.S. study finds that recent advances in AI have not caused a net decline in employment—rather than “killing jobs,” AI appears to be reshaping work by automating specific tasks while complementing others. Using granular employment data and firm- and local-level measures of AI adoption (such as AI-related software use, patenting, and shifts in advertised skill requirements), the researchers apply causal econometric techniques (difference‑in‑differences and instrumented exposure measures) to compare labor markets with high versus low AI uptake. The bottom line: aggregate employment and total hours remain largely stable, though firms and workers experience significant task reallocation and job turnover. For the AI/ML community, the findings matter because they emphasize augmentation over wholesale substitution. Technical implications include increased demand for models and interfaces that amplify human decision‑making, tools to streamline retraining, and systems that make AI outputs interpretable and integrable into human workflows. The study also highlights heterogenous effects—gains for AI‑complementary roles, disruption for narrowly routine tasks—underscoring the need for better AI adoption metrics, continued research on task-level impacts, and policy focus on reskilling and labor-market transitions rather than protectionism.