OhNoMyJob ↗RU

AI use and reported time savings

Time categories for selected AI-assisted tasks

Epoch / Ipsos, USA, 10–19 July 2026. Retrospective self-reports from employed or self-employed adults, in English. The ten knowledge-work activities are not a complete task or workday inventory.

Filters apply after “Show selected group”. Native source labels use English fallback. Work types are not exact SOC/ESCO occupations; groups overlap and must not be added together.

All selected valid task reports

Observation period: 2026-07-10 — 2026-07-19. Paired reports: 720; respondents: 284; selected task reports before exclusions: 798.

Weighted shares of selected valid paired task reports. Not shares of workers or saved hours. Scroll the table horizontally to see all columns.

Comparison of reported time categories
Current versus before AIShareSampling SE, ppReports, unweighted
Lower time category now39.4%2.82281
Same category; hours may differ44.2%2.68322
Higher time category now16.4%1.97117

SE clusters task reports by respondent. It covers sampling uncertainty only, not causality or future scenarios. Small cells are suppressed under the project reporting policy: at least 30 respondents, 5 events and 5 other reports. These are not risk thresholds or a source rule.

Excluded answers, interpretation and sources

Not sure: 55; new tasks, not a zero baseline: 21; undefined source −1 code: 2.

Respondents reported up to five selected AI-assisted tasks, prioritizing the tasks with most AI use. The closest-answer time bins include an open 20+ hour category. Equal bins can contain different hours. Selection, output, quality, review time and new work prevent interpreting these ordinal shares as causal hour replacement. No midpoint hours or detailed occupation/country allocation is inferred.

Epoch AI / Ipsos, Polling on AI usage, July2026: same-respondent task time before/current

2030 / 2035: this study does not provide a calibrated forecast of losing a job or of an involuntary reduction in paid hours that leaves insufficient income to live on. Risk categories and their four cutpoints remain unestablished.