AI Paradox: Promise vs. Reality—What It Means for Monetary Policy

Recent advances in artificial intelligence (AI) have revived expectations of transformative productivity gains and large-scale labour-market disruption. Yet despite rapid improvements in AI capabilities, aggregate productivity growth in advanced economies remains subdued, and widespread job displacement has not materialized. This divergence between technological promise and measured outcomes—the “AI productivity paradox”—poses important challenges for policy. This article synthesizes emerging empirical evidence on AI’s effects on labour markets and productivity. Near-term impacts are concentrated in within-occupation task restructuring and early-career hiring, while causal micro-level studies document sizable productivity gains (15–60 percent) that have yet to appear in aggregate statistics because of diffusion lags, organizational adjustment costs, and measurement limitations. We then examine the macroeconomic implications for potential output (Y*) and inflation dynamics. While AI is likely to boost potential output and exert disinflationary pressures over the long run, the effect on inflation during transition is much less certain. For monetary policy, the central challenge is distinguishing structural adjustment from cyclical weakness in real time. We argue that effective policy during the AI transition should exhibit measured flexibility.