Last year's AI models were good enough; the bottleneck is applications
MIT economist and Nobel laureate Daron Acemoglu on why automation alone fails, why Silicon Valley overinvests in foundation models, and what skills and businesses hold value as AI diffuses.