What the sources say
Acta — Can AI actually deliver the productivity gains it's promising?
, artificial intelligence has been celebrated as a once-in-a-generation productivity tool — yet the economic data has been slow to show the gains that advocates promised. The gap between AI's visible impact on individual tasks and its absence from macroeconomic productivity statistics is the central paradox this page examines.
Task-level productivity gains documented by controlled studies (MIT, Anthropic, Stanford)
of companies (6,000 CEOs/CFOs surveyed) report no impact on productivity or employment in the last 3 years
AI's estimated contribution to total factor productivity growth in 2025 — essentially zero
of enterprise AI pilots fail to generate meaningful ROI, per MIT NANDA study
US productivity growth in 2025 — just below the long-run average of 2%, far short of internet-era boom
Total productivity gain Daron Acemoglu (Nobel laureate) projects over the entire next decade