IMPLICATOR
OpenAI Chief Executive Sam Altman predicted in a July 25 Relentless podcast interview that AI would not fulfill technology’s promise of radically shorter workweeks at mass scale. OpenAI’s industrial-policy paper calls for incentives for employers and unions to test 32-hour schedules. He argued that rising expectations and economic competition would absorb the time saved by automation.
Key Facts
- Altman told the Relentless podcast on July 25 that AI won't bring a four-hour workweek at mass scale.
- OpenAI's April 6 policy paper proposes incentives for employers and unions to test 32-hour, four-day weeks without cutting pay.
- Gallup's July 27 report found 65% of employees at AI-adopting organizations saw productivity gains, but only 12% saw organization-wide change.
- A Nature Human Behaviour study of 2,896 employees found lower burnout and over 90% of companies kept four-day schedules after six months.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
Altman’s July forecast
Altman noted that technology had brought people more leisure and a higher quality of life, but had not fulfilled its most extreme promise. “But somehow we never get the promise of the four-hour workweek at mass scale in society,” he said. “And I don’t expect AI to change that.”
Altman said improving systems create more activity because expectations rise. He attributed that pattern to people comparing themselves with peers and wanting to remain useful.
OpenAI’s 32-hour proposal
OpenAI’s April 6 policy paper calls for incentives for employers and unions to run time-bound 32-hour, four-day trials without cutting pay. The recommendation appears in a section titled “Efficiency dividends.” The pilots would have to keep output and service levels constant.
When routine workloads decline and operating costs fall, the paper says employers could turn saved hours into a permanent shorter week, bankable paid time off or both. It presents the ideas as preliminary and does not announce an OpenAI workweek trial or a start date.
Gina Neff, a professor at the Minderoo Centre for Technology and Democracy, told BBC News the proposal would require “a complete change in the political headwinds to shift the balance between labour and capital.” Oxford Economics lead economist Adam Slater warned in a research note that many high-growth AI scenarios depend on optimistic assumptions about productivity and adoption, while gains from past technologies have taken decades to appear.
Gallup’s workplace data
Gallup’s July 27 report shows a gap between individual results and organizational change. Among employees at organizations that had implemented AI, 65% said it improved productivity and efficiency. Only 12% strongly agreed that it had changed how work gets done across their organization.
Stay ahead of the curve
Strategic AI news from San Francisco. No hype, no "AI will change everything" throat clearing. Just what moved, who won, and why it matters. Daily at 6am PST.
No spam. Unsubscribe anytime.
Gallup found that 13% of U.S. employees used AI daily and 15% used it a few times a week. Nearly half, 49%, said they never used it in their role. Gallup identified workflow integration and manager support as the strongest drivers of frequent use.
Know someone who'd find this useful? ✉️ Email it to a friend in one click, or they can subscribe free here.
Four-day-week trials
A study published in Nature Human Behaviour in July 2025 followed 2,896 employees. They worked at 141 organizations in six countries. Researchers reported lower burnout, better mental and physical health after six months, and more than 90% of the companies kept the schedule.
The study did not measure company-wide productivity. The participating companies volunteered, and employee outcomes were self-reported, so the authors called for randomized studies.
Sociologist Brendan Burchell’s 2023 account of an earlier British trial covered 61 organizations and about 2,900 workers. Seventy-one percent of employees reported less burnout, while sick days fell 65%. Average revenue increased 1.4% among the 23 organizations that supplied revenue data.
OpenAI’s April paper names no participants for the proposed pilots and gives no implementation timetable.
Frequently Asked Questions
What did Sam Altman say about AI and the workweek?
In a July 25 Relentless podcast interview, Altman said AI would not deliver the long-promised four-hour workweek at mass scale, arguing that rising expectations and economic competition absorb time saved by automation.
What is OpenAI's 32-hour workweek proposal?
OpenAI's April 6 industrial-policy paper, in a section called Efficiency dividends, calls for incentives for employers and unions to run time-bound 32-hour, four-day trials without cutting pay, provided output and service levels stay constant. It sets no start date.
What did Gallup's July 27 report find about AI at work?
Gallup found 65% of employees at AI-adopting organizations said it improved productivity, but only 12% said it changed how work gets done organization-wide. Just 13% of U.S. employees used AI daily, and 49% never used it in their role.
What did the four-day workweek study in Nature Human Behaviour show?
The July 2025 study followed 2,896 employees at 141 organizations in six countries, finding lower burnout and better health after six months. More than 90% of companies kept the schedule, though the study did not measure company-wide productivity.
What have economists and researchers said about OpenAI's proposal?
Gina Neff of the Minderoo Centre said the plan would need a complete change in the political headwinds between labour and capital. Oxford Economics' Adam Slater warned many AI productivity scenarios rest on optimistic assumptions, noting past technologies took decades to show gains.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
Marcus Schuler
Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: editor@implicator.ai
