AI productivity gains would tend to lower prices, but central bank responses can erase that effect; if AI leads to widespread labor displacement—higher unemployment alongside rising output—policymakers may respond with more monetary easing, potentially shifting the risk from deflation to inflation.

Main Finding

AI-driven productivity gains are intrinsically deflationary (via MV = PY), but sustained deflation in a modern economy would occur only if the central bank allowed it. With an operational Fed and its 2% inflation objective, the most likely outcome is that monetary expansion will offset AI’s price-downward pressure. A key risk instead is that large labor displacement from AI could raise unemployment even as output rises, prompting additional expansionary policy and shifting the greater risk from deflation toward inflation.

Key Points

• The marginal effect of productivity improvements (more Y) is deflationary if money (M) and velocity (V) do not adjust (MV = PY).
• Historical episodes (late 19th-century deflation and 1930s Great Depression) show money supply contractions or the absence/inaction of a central bank allowed deflation to persist; the later 1990s internet-era productivity boom did not produce deflation because the Fed expanded money.
• Modern central banking with an explicit inflation target makes sustained deflation from supply-side shocks unlikely unless policymakers fail to act or are constrained.
• Short-run factors (velocity shocks, credit contraction, policy lags) can generate temporary deflationary episodes (e.g., 2020 velocity drop), but the Fed can and has offset them via rapid monetary expansion.
• AI differs from past technology waves because it may displace labor even while raising output; empirical early evidence shows occupations with higher AI exposure experienced larger unemployment increases (Ozkan & Sullivan, 2025).
• If AI leads to persistent unemployment alongside rising output, the Fed’s dual mandate (price stability and maximum employment) could induce even more expansionary policy, raising inflation risk rather than causing deflation.
• Deflation remains possible in edge cases (central bank inaction, binding constraints like a gold standard analogue, severe financial frictions), and once deflation starts it can be self-reinforcing (deferred consumption, Fisher debt-deflation).

Data & Methods

• This is a conceptual/theoretical paper rather than an empirical study. Methods used include:
  • Monetary-arithmetic reasoning built on the equation of exchange (MV = PY) to show how increases in real output Y, holding MV fixed, imply lower P.
  • Historical comparison of episodes (1870–1900 classical gold-standard deflation; 1990s internet productivity gains; 1930s contraction) to illustrate the role of money supply and central-bank actions.
  • Literature synthesis and citation of recent empirical/structural work on AI and labor markets (e.g., Ozkan & Sullivan 2025; Wang & Wong 2026) and policy commentary (Barr 2026).
• No new micro- or macro-level data are introduced; conclusions rely on logical argumentation, prior empirical findings in the literature, and monetary theory.

Implications for AI Economics

• Policy focus: Central banks will likely respond to AI-driven supply shocks with monetary expansion to sustain inflation targets, meaning macro outcomes depend critically on policy reaction functions.
• Labor–output decoupling: If AI causes persistent labor displacement while output rises, macroeconomic models should explicitly allow for simultaneous increases in output and unemployment; standard Phillips-curve intuitions may break down.
• Modeling and forecasting: Researchers should incorporate two channels into forecasts: (i) direct productivity effects on prices via MV = PY, and (ii) second-round macro effects through employment, consumption, and monetary policy responses.
• Risk assessment: The larger macro risk may shift from deflation to inflation if policymakers pursue aggressive offsetting measures in response to labor-market deterioration; scenario analyses should evaluate both deflationary and inflationary tails depending on central bank constraints and credibility.
• Historical lessons: Work on AI economics should account for institutional constraints (e.g., monetary regime, central-bank capacity, financial frictions) because they condition whether productivity-driven price declines materialize.
• Empirical priorities: Collecting high-frequency evidence on AI exposure by occupation, its realized effects on employment and wages, and central-bank reaction functions will be crucial for assessing likely macro outcomes.