Why The 15-Hour Week Never Arrived In The Age Of AI
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

John Maynard Keynes predicted in 1930 that living standards in wealthy countries could rise four to eight times within a century; estimates cited by Thorsten Meyer AI put the increase at about 4.4 times in Britain and 5.5 times in the United States by 2022. But Keynes described a 15-hour week as one way to share remaining work, not as a firm deadline, and U.S. factory hours fell to around 40 a week before leveling off. Recent evidence cited in the source suggests generative AI saves some work time, while a Danish study found no measurable change in recorded hours.

Thorsten Meyer AI’s November 18 episode examines why the 15-hour workweek John Maynard Keynes discussed in 1930 did not become the norm, despite large gains in living standards. The episode’s account says U.S. manufacturing hours fell sharply before 1940 and then stayed near 40 hours a week, while current evidence on generative AI has not yet shown a broad reduction in recorded working time.

Before you orderOffer from Amazon

Get everyday essentials delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

Keynes’s essay, “Economic Possibilities for our Grandchildren,” predicted that living standards in “progressive countries” might become four to eight times higher within a hundred years. Meyer AI calculates that by 2022 real income per person had reached about 4.4 times its 1930 level in Britain and 5.5 times in the United States, using data from the Maddison Project. The figures are the episode’s calculations, not a claim that Keynes’s other expectations came true.

The source stresses that Keynes did not set a deadline for a 15-hour week. He suggested “three-hour shifts or a fifteen-hour week” as a way to distribute the work that remained if productivity reduced the need for labor. U.S. manufacturing production workers averaged 51 hours a week in 1909, 44.2 in 1929, and 41.8 in October 2026, according to figures cited from the Census Bureau’s historical statistics and current data in the episode.

The Fair Labor Standards Act established a 40-hour threshold for overtime pay starting in October 1940: hours beyond it generally required pay at one and a half times the regular rate. Meyer AI describes 40 hours as the standard American week since then. For recent AI evidence, the episode estimates that generative AI saved U.S. workers about 2.1% of all work hours in August 2026. It also cites a Danish study that found no measurable effect of AI chatbots on recorded hours.

At a glance
analysisWhen: Published November 18, 2026; AI work-ho…
The developmentA November 18, 2026, Thorsten Meyer AI episode revisits Keynes’s 1930 essay and compares its productivity hopes with the persistence of the roughly 40-hour workweek and early evidence on AI.
Work and AI · Briefing

Why the 15-Hour Week Never Arrived in the Age of AI

Living standards rose sharply, but the typical U.S. workweek did not shrink in step. Early AI evidence shows time savings without clear proof of shorter schedules.

2.1%
estimated U.S. work hours saved by generative AI in August 2026
4.4×
Britain’s real income per person by 2022 versus 1930
5.5×
U.S. real income per person by 2022 versus 1930
15 hours
one possible way Keynes suggested sharing remaining work

A prediction, not a deadline

Keynes forecast living standards could rise four to eight times in wealthy countries; his 15-hour week was a proposal for distributing work if productivity reduced labor needs.

What Keynes predicted
  • Living standards four to eight times higher within a century
  • Three-hour shifts or a 15-hour week as possible ways to share remaining work
What followed
  • Income per person reached about 4.4× in Britain and 5.5× in the United States by 2022
  • The 15-hour week did not become the norm

Factory hours fell, then leveled off

U.S. manufacturing production workers moved closer to 40 hours a week, while overtime law helped establish that threshold.

1909
Workers averaged 51 hours a week.
1929
The average had fallen to 44.2 hours.
October 1940
The 40-hour overtime threshold took effect.
October 2026
Manufacturing production workers averaged 41.8 hours.

AI time savings have not settled the question

Saved task time can become more output or additional work instead of shorter schedules.

U.S. estimate

Some time saved

The episode estimates generative AI saved workers about 2.1% of all work hours in August 2026.

Danish study

No measurable hours change

The cited study found no measurable effect of AI chatbots on recorded hours.

What would show a lasting shift?

Evidence needs to connect task efficiency with how people actually work.

  • Track whether AI time savings persist across occupations.
  • Compare recorded hours, work schedules, and compensation over time.
  • See whether employers and workers use freed time for leisure, more output, or both.
CaveatThe 2.1% estimate does not establish shorter schedules, and the supplied material lacks details about the Danish study’s sample and measurement period.
The takeaway
Higher productivity can raise living standards without reducing work time; AI’s effect on the workweek depends on how its gains are shared.

Why Productivity Has Not Cut Hours

The gap between higher output and shorter working time matters because productivity gains can be used in different ways: to raise income, expand production, or create more leisure. The figures in the episode suggest that rising living standards alone did not lead to a comparable fall in the typical U.S. workweek. The 40-hour threshold became embedded in law and workplace practice, while the episode’s evidence does not show that recent AI adoption has yet shifted that pattern at scale.

The AI estimates are early indicators, not proof of a long-term trend. A reported reduction in time spent on tasks does not automatically mean employees work fewer hours; organizations may use saved time for additional tasks or output. Whether AI produces more leisure will depend on how employers, workers, and policymakers distribute its gains.

Products Worth Considering

As an affiliate, we earn on qualifying purchases.

From Factory Hours to Forty

Keynes wrote during the onset of the Great Depression, when he argued that widespread pessimism underestimated the potential for economic progress. He expected compound interest and scientific advances to ease the struggle for subsistence. He also warned that people would still want some work and raised the question of how they would use greater leisure.

The working week was already declining in U.S. manufacturing: the episode gives averages of 51 hours in 1909, 49.4 in 1914, and 45 in 1926. It notes that Ford cut its working day to eight hours in 1914 and announced a five-day, 40-hour week for its plants in 1926. The Fair Labor Standards Act, signed in 1938, phased the overtime threshold from 44 hours to 42 and then 40 by October 1940.

Germany later saw a separate reduction in metalworking hours. According to the source, a 1984 settlement after a seven-week IG Metall strike set a 38.5-hour week, and the West German metal industry reached 35 hours by 1995. These examples show that working hours can change through labor agreements and policy, though outcomes differ by country and sector.

“Generative AI saved U.S. workers about 2.1% of all work hours.”

— Thorsten Meyer AI episode, summarizing its estimate for August 2026

What AI Hours Data Can Show

The source does not provide the Danish study’s full citation, sample, or measurement period, so the finding cannot be assessed in detail here. The 2.1% time-saved estimate also does not specify in the supplied material how the saved time was measured across jobs or whether it translated into shorter schedules. It is not clear whether AI will reduce total hours, increase output, or do both in different workplaces.

The 40-hour figure is a broad benchmark, not a description of every worker’s schedule. The source’s 41.8-hour figure applies specifically to U.S. manufacturing production workers in October 2026. It does not establish the average for all U.S. workers or explain differences by occupation, pay, or employment status.

Whether AI Changes the Workweek

The next evidence to watch is whether measured AI time savings persist and show up in workers’ recorded hours, schedules, or compensation. Longer-running studies across occupations would help distinguish short-term task efficiency from lasting changes in work time. The episode does not identify a scheduled policy decision or a specific next milestone; for now, the central question is how employers and workers use time freed by automation.

Key Questions

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

No. Keynes offered a 15-hour week as a possible way to share remaining work. The episode distinguishes that suggestion from his numerical prediction about future living standards.

How did living standards compare with Keynes’s prediction?

Meyer AI 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, within Keynes’s predicted range of four to eight times for “progressive countries.”

Why did the U.S. workweek level off near 40 hours?

The episode points to the Fair Labor Standards Act, whose 40-hour overtime threshold took effect in October 1940, alongside the leveling off of manufacturing hours. The source does not establish one cause for the broader pattern.

Has generative AI reduced working hours?

The evidence cited is mixed and limited. Meyer AI estimates that generative AI saved U.S. workers about 2.1% of all work hours in August 2026, while a Danish study cited in the episode found no measurable effect on recorded hours. Saved task time does not necessarily mean shorter workweeks.

Source: Thorsten Meyer AI

You May Also Like

Apple greift nach China-Speicher. Europa hat nicht einmal diese Option.

Apple plant, Speicherchips vom chinesischen Hersteller CXMT zu beziehen, während Europa keine vergleichbaren Optionen hat. Das zeigt die Abhängigkeit Europas von asiatischer Speicherproduktion.

The Eye Over the City: How Wide-Area Motion Imagery Works — and Where It Goes Blind

An in-depth look at WAMI technology, its capabilities, limitations, and future applications in surveillance and defense.

Top European AI Suppliers For 2026: Which Companies Stand Out?

An analysis of leading European AI companies in 2026, highlighting ownership, certification, and strategic importance for European sovereignty.

Trump presiona a Cuba y planea enviar a Uruguay a los deportados cubanos

Former President Trump is reportedly pressing Cuba and planning to send deported Cubans to Uruguay, raising diplomatic and immigration concerns.