Repaying the inheritance: How education and research policy can address AI's borrowed expertise

This Brookings Institution article argues that the current productivity boom from generative AI relies on expertise built before AI tools existed, creating an 'inheritance' that is being spent without adequate replenishment. The author warns that juniors are doing less developmental work, firms hire fewer of them, and research culture is shifting toward recombination rather than deep expertise. The piece compares the challenge to the calculator's introduction in education, noting three ways AI is harder: AI can be confidently wrong, it frames problems rather than just executing operations, and the convenience differential is much larger. The author calls for schools, universities, professional bodies, and funders to develop vocabulary and instruments to manage the tension between short-term productivity and long-term capability building, before decisions are made by default.