Pascual Restrepo is an economist who studies automation, AI, economic growth, and labor markets. He has coauthored research with Daron Acemoglu and serves on Anthropic's economic advisory council. In an August 2026 interview with The Bell, he described AI not as an instantaneous labor-replacement shock but as a transition constrained by uneven model capabilities, slow organizational adoption, energy, and scarce inference compute.source: the-bell-pascual-restrepo-ai-labor-economy-2026.md
Restrepo's distinctive argument links wages to compute opportunity cost. If an AI can replace a worker but the required tokens could generate greater value in science, medicine, or another scarce domain, employing the human may remain economically rational. As compute supply expands, more tasks cross the automation threshold; labor's share of output can therefore shrink dramatically even while real wages and purchasing power remain above today's levels because AI also lowers prices across the economy.source: the-bell-pascual-restrepo-ai-labor-economy-2026.md
He expects the transition to produce concentrated losers before general-purpose price declines arrive. His policy priorities are real-time measurement of sector-level "pockets of disruption," social insurance for workers whose accumulated human capital is suddenly devalued, continued datacenter construction with subsidies for socially valuable compute, and separate safety certification for frontier models. This connects the distributional problem in ai-labor-market to the allocation problem in ai-compute-infrastructure.source: the-bell-pascual-restrepo-ai-labor-economy-2026.md
Related pages: ai-labor-market, ai-compute-infrastructure, dead-economy-theory, ai-job-grief, anthropic, ai-usage-telemetry.