2.1%
AI assisted 6.3% of US work hours and saved 2.2% in May 2026, then 5.3% and 2.1% in August 2026; the hours series start at 4.1% assisted in November 2024.
A program of EffectiveSC
How We Use AISources checked October 24, 2026
Where things stand / Work and productivity
Time saved, work hours assisted, and what experiments and companies report.
Most time-saved figures rest on what workers report. The St. Louis Fed spreads them over every hour worked in the US, by users and non-users alike.
2.1%
AI assisted 6.3% of US work hours and saved 2.2% in May 2026, then 5.3% and 2.1% in August 2026; the hours series start at 4.1% assisted in November 2024.
about 5%
Project calculation: dividing hours saved by the adoption rate gives about 5% of work hours saved for the average adopter (2.08 / 40.55 in August 2026).
about 3%
Two years after ChatGPT's launch, Danish register data rule out effects on earnings or hours above 2%; adopters report saving about 3% of work hours.
Controlled experiments measure output instead of asking people. Results depend on the task and the worker: some groups gain a lot, and some do worse.
15%
Across 5,172 customer-support agents, an AI assistant raised issues resolved per hour by 15% on average (QJE, 2025); the 2023 working paper gives 5,179 agents and 14%.
36%
The least-skilled support agents gained about a third: 36% more issues resolved per hour for the lowest skill quintile (QJE; 34% in the working paper). The journal's summary gives about 30% for less-skilled and less-experienced workers.
1.4 hours
In a Microsoft-run experiment across 66 firms and 7,137 knowledge workers, those offered Copilot spent 1.4 fewer hours a week on email (12%); those who used it saved about 2 hours (17%). Meeting time didn't change.
Disclosure: Microsoft ran the experiment, and some of the authors work at Microsoft Research.
19%
In METR's early-2025 randomized trial, 16 experienced open-source developers took 19% longer with AI tools, yet afterward believed AI had saved them 20%.
METR has since said newer data suggest a speedup, but it calls that evidence weak because of who took part.
nearly 10%
A GPT-4 business assistant on WhatsApp had no average effect on 640 Kenyan small businesses: owners below the median did nearly 10% worse, while those above it may have gained over 15% (not statistically significant).
43%
On tasks inside AI's frontier, below-average consultants improved 43% with GPT-4 against 17% for above-average ones; on a task just outside it, consultants were right 60% to 70% of the time with AI against about 84% without.
Part of the time AI saves goes back into checking and fixing its output. The estimates come from worker surveys, not measured time.
40%
Nearly 40% of the time AI saves is lost to rework, and only 14% of employees consistently get clear, positive net outcomes.
Who was asked: 3,200 full-time employees who use AI
Disclosure: A survey by Workday, a software vendor.
4 hours
US workers say AI saves them about 6 hours a week, but they spend about 4 hours a week fixing its output.
The hours spent fixing may cover only the workers who report reviewing AI's output.
Most executives report no effect on productivity so far. Their figures are their own estimates, not company accounts.
83%
In mid-2026 surveys of nearly 9,000 executives in seven countries, 78% of firms used some AI (95% in the US), yet 83% reported no effect on labor productivity over the past three years.
The US share leaves out firms that skipped the question.
1.8%
CFOs said AI raised their firms' 2025 labor productivity 1.8% on average, but their own estimates of AI's effect on revenue and headcount imply 0.6%.
See also: Are companies getting a return on AI?
The AI at Work Benchmark
Score your organization's practices, measure your people with an anonymous pulse, and see both against the published evidence.