AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

A November 18, 2026, essay revisits economist John Maynard Keynes’s 1930 discussion of shorter working hours and compares it with current data. It says US factory workers averaged 41.8 hours a week in October 2026, while separate early evidence on generative AI and work hours remains mixed.

A November 18, 2026, report revisits John Maynard Keynes’s 1930 discussion of shorter working hours and finds that, despite large gains in income and new tools such as generative AI, the standard US workweek remains close to 40 hours. The piece says US factory workers averaged 41.8 hours a week in October 2026, while early evidence on AI’s effect on work time varies by measure and study.

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

Keynes predicted that living standards in what he called “progressive countries” would rise to between four and eight times their 1930 level over the following century. The report calculates that by 2022, real income per person was about 4.4 times its 1930 level in Britain and 5.5 times in the United States, using Maddison Project data. Those figures address income, not how many hours people work.

The report traces a sharp fall in US manufacturing hours before the middle of the 20th century: production workers averaged 51 hours a week in 1909, 44.2 in 1929, and 41.8 in October 2026. It identifies October 1940, when the Fair Labor Standards Act’s overtime threshold reached 40 hours, as the point when that figure became the standard American workweek. The law did not prohibit longer weeks; it required overtime pay above the threshold.

On AI, the report cites an estimate that US workers saved about 2.1% of all work hours with generative AI in August 2026. It also points to a Danish study that found no measurable effect of AI chatbots on recorded hours. These findings measure different settings and do not establish that AI has shortened the typical workweek.

At a glance
reportWhen: Published November 18, 2026; includes U…
The developmentA November 18 essay examines why rising incomes and AI-driven time savings have not produced a shorter standard working week.
Work and AI · Briefing

Why AI Hasn’t Made Workdays Shorter

Task-level time savings have not yet shown up as a shorter standard workweek. The early evidence is mixed, and laws and bargaining have helped shape working hours.

41.8 hours
US factory workers’ weekly average in October 2026
2.1%
Estimated work hours saved with generative AI in August 2026
40 hours
US overtime threshold reached in October 1940

Income gains did not set the schedule

By 2022, real income per person had risen substantially since 1930, but income growth alone did not shorten the standard workweek.

CountryReal income per person by 2022
BritainAbout 4.4 times its 1930 level
United StatesAbout 5.5 times its 1930 level

The long decline in factory hours

US manufacturing hours fell over time, while the 40-hour overtime threshold became a widely used benchmark.

1909
Production workers averaged 51 hours a week.
1929
The weekly average was 44.2 hours.
October 1940
The federal overtime threshold reached 40 hours; longer weeks remained possible with overtime pay.
October 2026
US factory workers averaged 41.8 hours a week.

What the AI evidence shows

The cited findings measure different things, so they do not establish that AI has shortened typical weekly schedules.

Time saved
  • One estimate says US workers saved about 2.1% of all work hours with generative AI in August 2026.
  • The report does not provide enough detail here to assess the estimate’s methods.
Recorded hours
  • A Danish study found no measurable effect of AI chatbots on recorded hours.
  • The report does not give the study’s sample size or observation period.

Institutions can shape working time

Historical examples show that employers, laws and collective bargaining can turn changing conditions into different schedules.

1
Company changes

Ford moved from a nine-hour to an eight-hour day in 1914, then announced a five-day, 40-hour week for its plants in 1926.

2
Federal threshold

The Fair Labor Standards Act phased the overtime threshold from 44 to 42 to 40 hours.

3
Collective bargaining

After a seven-week strike in 1984, a German metalworking settlement reduced weekly hours to 38.5; western Germany’s industry reached 35 hours by 1995.

What remains unclearThe figures do not show whether AI time savings became fewer scheduled hours, more output or different tasks. The factory-hours figure covers manufacturing, not all US workers.
The takeaway
AI may save time on tasks, but a shorter workweek depends on whether those gains are converted into fewer hours.

Why Time Savings Have Not Cut Hours

The gap between time saved on tasks and hours worked in a week matters to workers deciding what new technology means for their schedules. A tool may help complete particular tasks faster, but that does not by itself show that employers or workers have reduced total hours. The report’s figures suggest that productivity gains and rising income have not automatically changed the widely used 40-hour benchmark.

The evidence offered here is an early snapshot, not a complete account of how AI affects work across occupations. The US estimate measures hours saved with generative AI, while the Danish study examines recorded hours after chatbot adoption. Neither figure alone explains how saved time is distributed among workers, employers, or additional tasks.

Products Worth Considering

Kindle Paperwhite Signature Edition 32GB (2024 Model) - Metallic Black

Kindle Paperwhite Signature Edition 32GB (2024 Model) – Metallic Black

As an affiliate, we earn on qualifying purchases.

From Keynes’s Forecast to the 40-Hour Norm

In his 1930 essay “Economic Possibilities for our Grandchildren,” Keynes argued that compound growth could ease the struggle for subsistence. He wrote that people would still need some work for satisfaction, and proposed “three-hour shifts or a fifteen-hour week” as one way to share the remaining work. The report stresses that the 15-hour week was not a dated prediction; Keynes presented it as a possible arrangement for distributing work.

The historical decline in US manufacturing hours gave that possibility some basis. Ford reduced its working day from nine hours to eight in 1914, according to the report. In 1926, the company announced a five-day, 40-hour week for its plants. Federal law later set a phased overtime threshold: 44 hours in the first year, 42 in the second, and 40 from October 1940.

The report also notes a later European example. Following a seven-week strike in 1984, Germany’s metalworking settlement reduced weekly hours to 38.5; the industry’s working time in western Germany reached 35 hours by 1995, according to IG Metall. These examples show that policy and collective bargaining can affect working time alongside changes in technology.

““Five working days of eight hours, with no overtimes, is to constitute a week’s work.””

— Ford company paper, quoted in the report on the 1926 announcement

What the AI Hours Data Cannot Show

The report does not provide enough detail here to assess the methods behind the 2.1% time-savings estimate, including its sample, occupations, or calculation. It also does not give the Danish study’s sample size, observation period, or full results. The reported findings therefore do not show whether time saved by AI led to fewer scheduled hours, more output, or different tasks.

The latest factory-hours figure is a sector-specific measure, not a count of hours worked by every US employee. The source also does not establish whether the October 2026 figure is a monthly average or how it compares with other worker groups. The relationship between AI use and total working time remains unsettled in the material provided.

The Next Test Is Total Hours

Further evidence will need to track AI adoption and total hours over time, with enough detail to distinguish time saved on individual tasks from changes in weekly schedules. Comparisons across occupations and countries could show whether the early estimates apply beyond the groups and periods measured so far.

The report’s historical comparison also points to institutions as part of the story: laws and bargaining helped define working-time standards in the United States and Germany. Whether AI-related productivity gains lead to shorter weeks will depend on how those gains are measured and whether employers, workers, or policymakers convert them into additional leisure.

Key Questions

Did Keynes predict that people would work only 15 hours a week by 2030?

No. The report says Keynes’s 15-hour week was a possible way to share the remaining work, not a dated forecast. His specific numerical forecast concerned living standards over a century.

How many hours did US factory workers work in October 2026?

The report gives an average of 41.8 hours a week for US production workers in manufacturing in October 2026. That is a sector-specific figure, not a measure for all US workers.

Has generative AI reduced the standard workweek?

The evidence cited does not establish that. One estimate says US workers saved about 2.1% of all work hours with generative AI in August 2026, while a Danish study found no measurable effect of AI chatbots on recorded hours.

When did 40 hours become the standard US workweek?

The report dates the standard to October 1940, when the Fair Labor Standards Act’s overtime threshold reached 40 hours. The law required overtime pay above that threshold; it did not ban longer weeks.

Source: Thorsten Meyer AI

You May Also Like

What Changes When AI Makes Intelligence Less Scarce?

Thorsten Meyer’s 75,000-word book on AI, work and economic power entered ebook distribution on October 16, 2026.

Cross-platform buyer history for multi-marketplace resellers

Resellers selling across eBay, Poshmark, and Mercari are testing a manual buyer ledger to unify buyer history, aiming to improve pricing and customer management.

The Management Tells: What Five Frontier AIs Reveal Under Pressure

Can you identify an AI by how it manages a crisis? Firmulate turns 242 real decisions into a revealing test of judgment and follow-through.

Show HN: Fine-tune An 8B Model On A 4 GB Laptop GPU

A developer demonstrates fine-tuning an 8-billion-parameter AI model using only a 4 GB GPU, challenging assumptions about hardware requirements for large models.