AI users report time savings more often than better quality
Kristen E. Broady · Federal Reserve Bank of Chicago · 2026-11
Source finding
Among workers who used generative AI at work in the prior month, weighted agreement is 81.0 percent for time savings and 51.8 percent for better quality. Two separate questions, not parts of a whole.
Our interpretation
Perceived time savings do not establish quality. Assess both measures and business results separately.
Method and evidence
Kristen E. Broady analyses weighted SHED 2025 data. 7,488 workers answered the usage question; 1,850 reported AI use. Table 1, Panel C, describes these users. Observational self-report; moderate evidence for reported experiences.
Limitations and interests
Unreviewed working paper, no causal test or objective assessment of time and quality. Published by the Federal Reserve Bank of Chicago; views are the author’s, not an official position. Publication page: November 2026; PDF: October 2026. No exact day established.
What businesses can do
Measure time saved and result quality separately. Check work quality and rework before treating time savings as business value.
Where to start
Table 1, Panel C and discussion on PDF pages 17–19. Open original report.