METR_Evals survey finds AI tools raise output value 1.6 to 2.1 times
METR researcher Joel Becker posted results from a METR_Evals survey of 349 technical researchers, engineers and managers conducted between February and April 2026. Respondents said AI tools at work multiply the value of their output by an average factor of 1.6 to 2.1, with the multiplier reported as increasing over time. The findings rest on self-reported productivity assessments from participants in research, engineering and management roles.
#389Erik Brynjolfsson@ERIKBRYNMore evidence that AI is boosting productivity, this time from @joel_bkr and the team @METR_Evals
METR@METR_EvalsWe surveyed 349 technical researchers, engineers, and managers (in February–April 2026) about how they use AI tools at work. On average, participants self-report that AI use made their work 1.6–2.1x more valuable, and that this multiplier will grow over time.
#918Zvi Mowshowitz@THEZVIThis is a pretty crazy low expectation from software engineers, that they went from 1.3x speedup to 2x speedup in a year and then expect this will slow down in absolute terms to 2.5x in another year. You'd get to 2.5x just from learning to better use existing AI.
#1389Jaime Sevilla@JSEVILLAMOLThere is no way that SWEs are being sped up 2x, but I can believe there is a very solid productivity increase already in place, and that it will become even larger in the future.
Joel Becker@joel_bkrnew research from me @METR_Evals: technical workers claim that today's AI impacts value of their work to an extraordinary degree (& growing over time). of course, self-reports plausibly overestimate. the magnitudes nonetheless strike me as remarkable.
#1553Joel Becker@JOEL_BKRnew research from me @METR_Evals:
technical workers claim that today's AI impacts value of their work to an extraordinary degree (& growing over time).
of course, self-reports plausibly overestimate. the magnitudes nonetheless strike me as remarkable.
METR@METR_EvalsWe surveyed 349 technical researchers, engineers, and managers (in February–April 2026) about how they use AI tools at work. On average, participants self-report that AI use made their work 1.6–2.1x more valuable, and that this multiplier will grow over time.