Science & Tech · AI · 5 hrs ago
Human checks may be reducing the efficiency gains from AI
Companies have promoted AI as a way to save time and make office work more efficient.
But a survey of nearly 6,000 senior executives found that about 90% saw no effect on jobs or productivity, even though many companies were using AI.
One reason may be the time workers spend checking whether AI results are accurate, safe and usable.
A Workday survey of 32,000 employees and business leaders found they used an average of 37% of the time saved to fix or rewrite poor results.
Google researchers also found that scientists spent some of their time savings checking AI output, while many still had more ideas awaiting tests.
This review work can create a bottleneck and reduce the efficiency AI is meant to deliver.
The sources do not say what happens next.
Many companies report little or no productivity gain from artificial intelligence, despite actively using it.
A major reason may be the human work needed to check and correct AI-generated results.
A Workday survey found that workers used about 37% of time saved with AI to fix, clarify or rewrite poor-quality outputs.
A survey of nearly 6,000 senior executives found about 90% reported no effect from AI on employment or productivity.
Research also suggests AI may shift bottlenecks to the work of testing and verifying the ideas it generates.
- Who
- Workers and companies using AI, as well as researchers studying its effects.
- What
- Human review and correction of AI output may be reducing the technology’s efficiency gains.
- When
- The article was published on 2026-10-12. The studies it discusses were conducted or published at different times.
- Where
- The studies covered in the article include research by organizations in the United States and Mexico. The Workday survey covered employees and business leaders.
- Why
- AI-generated results often need to be checked, corrected or rewritten, taking up some of the time the technology saves.
This story does not have two clearly opposing sides.
Christian Catalini wrote in a working paper that human verification capacity was a constraint on growth.
Researchers from the Atlanta Federal Reserve, Stanford University, the Bank of England and the Autonomous Technological Institute of Mexico published a survey of nearly 6,000 senior executives.
Apollo chief economist Torsten Slok said AI’s productivity return remained a forecast rather than a finding.
- Executive survey
- Nearly 6,000 senior executives
- Reported no impact
- About 90% said AI had no effect on employment or productivity
- Active AI use
- 69% said their companies used AI
- Time spent correcting AI
- About 37% of time saved, on average
- Scientist interviews
- 600 scientists interviewed by Google economists
What people are saying
The return of AI in terms of productivity remains a forecast, not a finding.
The main constraint on growth is no longer intelligence.
The only useful AI is AI that is verified.
Basically, we found that it seems the bottlenecks are now shifting downstream.
Sources
Related news
Microsoft president calls for global AI safety rules
Meta launches Muse Wonder models to compete with leading AI systems
Anthropic AI submits false murder tip to Philadelphia police
CEATEC 2026 opens in Chiba with 821 exhibitors
Meta unveils two AI models it says rival leading competitors
Scholar urges Global South to move beyond supplying AI raw materials
Meta launches new AI models, with faster development planned
Robotics competitions in Hangzhou to feature embodied AI challenges
Samsung outlines AI memory strategy beyond faster HBM
Meta launches Muse models, with Muse Wonder ranked second in performance evaluation
Anthropic says new safeguards sharply reduce AI distillation attacks
