Bot the difference: AI’s absence in economic data

What It Covers

The Economist's Intelligence examines why AI's rapid capability growth has not yet appeared in U.S. economic productivity data, using 2025 macroeconomic figures and three-variable analysis of adoption rates, usage intensity, and measured efficiency gains to estimate AI's actual current contribution to worker output.

Key Questions Answered

• •AI Productivity Calculation: Estimate AI's real economic contribution using three variables: adoption rate (40% of U.S. workers), usage intensity (average 2 hours per week, roughly 5-6% of working hours), and measured efficiency gains (15-30% per task). Combining these yields a productivity increase of only 0.25-0.5 percentage points — likely an overestimate given real workplace behavior.
• •Usage Intensity Gap: Despite 40% of working-age Americans using AI on the job, only 13% use it daily. This low intensity severely limits aggregate productivity impact. Analysts and investors tracking AI's economic footprint should weight frequency of use alongside adoption figures, as headline adoption numbers significantly overstate actual productive deployment.
• •Productivity Redeployment Problem: Efficiency gains from AI do not automatically convert into economic output. Research on tech workers shows that time saved through AI gets redirected into longer hours and experimentation rather than additional productive output. Firms and policymakers should not assume AI time savings translate directly into GDP growth without measuring how freed time is actually redeployed.
• •Historical Reorganization Pattern: Productivity booms from general-purpose technologies — electricity, computers, now AI — occur when firms redesign operations around the technology, not when workers simply adopt new tools. The electricity analogy is instructive: gains arrived when factory floor plans were restructured, not when motors replaced steam engines. Business model transformation, not tool adoption, drives measurable productivity gains.
• •GDP-Employment Gap Context: The 2025 U.S. data showing 2.2% real GDP growth alongside only 0.1% employment growth appears anomalous but is historically common — in one-third of years since 1950, this gap exceeded two percentage points. The 2025 gap also reflects AI infrastructure investment inflating output and immigration policy reducing lower-productivity worker counts, not a genuine productivity breakthrough.