Measuring the Self-Reported Impact of Early-2026 AI on Technical Worker Productivity

Doing tasks faster and producing more valuable work are different productivity claims. METR's survey of technical workers asks about both because AI can change which tasks people choose. A newly cheap task may save many hypothetical hours without contributing as much to the project as the work it displaces.

Participants reported greater speed gains than value gains, but both are perceptions, not experimentally measured effects. The convenience sample and low response rates also limit generalization. Asking several versions of the question helped check whether answers were internally consistent; it could not establish that respondents accurately knew what their work would have looked like without AI.