CEOs Report AI Productivity Gains Falling Short

Global reporting aggregates several recent surveys showing widespread executive frustration with AI's near-term productivity effects. Global News, citing a Globalization Partners survey of 2,850 business executives, reports 73% of respondents called returns from AI investments "underwhelming," and 70% said they want to see productivity gains this year or will scale back AI spending if gains do not materialize (Global News, May 12, 2026). The Globalization Partners data also found 88% of respondents worried employees use AI to appear productive and 69% said time spent monitoring or updating AI outputs has increased (Global News).

Fortune summarizes a National Bureau of Economic Research (NBER) analysis covering roughly 6,000 executives across the U.S., U.K., Germany, and Australia that found most firms report little impact from AI on employment or productivity; the NBER work also reports typical AI users log about 1.5 hours per week of AI use (Fortune). Inc.'s April survey of 2,400 knowledge workers and leaders found 64% of company leaders interact with AI at least two hours daily, yet 48% of executives called AI adoption a "disappointment," with 29% reporting significant ROI from generative AI and 23% reporting significant ROI from more autonomous systems. Inc. also reports widespread concerns about data leaks and internal conflict tied to AI rollouts (Inc., Apr 7, 2026).

These survey results fit a recurring empirical pattern in technology adoption, historically summarized by the Solow productivity paradox, where new IT does not immediately show up in macro productivity statistics. Industry-pattern observations: early-stage AI deployments frequently produce measurement gaps because benefits are diffuse, accrue over longer time horizons, and are offset by increased oversight, quality control, and integration work. Practitioners typically see a mix of surface-level task acceleration and hidden costs-time spent reviewing, validating, and correcting model outputs-that can reduce net productivity gains in the short run.

Reporting highlights two tensions for organizations: heavy executive investment and expectation-setting on one hand, and operational friction and governance challenges on the other. Fortune notes executives still forecast modest productivity gains in coming years even while reporting limited current impact (Fortune). Inc.'s findings add behavioral and cultural frictions-stress among leaders, fears of job risk, and reports of tools generating internal power struggles and data leakage concerns (Inc.). Industry-pattern observations: when technology adoption outpaces governance and measurement frameworks, organizations commonly see a period of rework and reorganisation before stable gains appear.

• •Usage intensity vs. outcome metrics: whether reported hours of AI use translate to measurable throughput or quality improvements.
• •Monitoring overhead: trends in time spent reviewing or correcting AI outputs versus time saved automating tasks.
• •Governance incidents: frequency of data leaks or misuse tied to unapproved tools, as flagged in Inc.'s survey.
• •Procurement signals: whether a meaningful share of executives follow through on scaling back budgets if near-term gains remain absent (Global News).

For practitioners: these sources suggest prioritizing rigorous A/B measurement, tighter validation workflows, and clearer metrics linking AI activities to business outcomes before declaring success.

(Reporting sources: Global News summary of a Globalization Partners survey; Fortune summarizing NBER analysis; Inc. coverage of a knowledge-worker survey.)