AI may lift wages — then crush them, professors say (www.businessinsider.com)

A new Brookings Institution paper by Konrad Kording and Ioana Marinescu (University of Pennsylvania) uses an interactive simulation of a transition from human-led to machine-led “intelligence” work to argue wages will follow a hump‑shaped path: early AI-driven productivity gains raise pay as humans work with smarter tools, but as automation spreads and AI masters more cognitive tasks, demand for those human workers falls and wages decline. Crucially, output keeps rising even as wages slip — implying the benefits increasingly accrue to capital (machines and owners) rather than labor — and many displaced workers are pushed into lower‑value, slower‑growing physical jobs, creating a long‑term drag on wages. For the AI/ML community this reframes adoption choices as distributional as well as technical: models and deployments that substitute for human expertise can produce short‑term wage boosts but risk “intelligence saturation,” where further gains slow because human and physical bottlenecks remain. The authors recommend policy responses — slowing automation’s pace, investing in physical capital so human labor stays productive, and taxing virtual substitutes (a “robot tax”) — to prevent concentrated gains and wider labor dislocation. The paper underscores that design, rollout timing and complementary investments will determine whether AI raises broad prosperity or chiefly enriches capital holders.