Toto-Rex's story
Toto-Rex's company rolls out an AI assistant. A report that took a day now takes an hour.
So Toto-Rex's manager adds three new projects. Toto-Rex answers messages at dinner. Then at midnight.
Toto-Rex's partner says: 'You told me AI would give you your evenings back.' 💔 Toto-Rex's story is made up. The studies in this video are real.
The short answer
Not automatically.
The details
Researchers observed one US tech company for eight months.
At that company, staff using AI worked faster and took on more tasks. They worked into more hours of the day.
Often, nobody asked them to.
A separate Danish study looked at about 25,000 workers in 11 AI- exposed occupations.
Using workers' reports, the researchers estimated time saved on tasks equal to about 3% of chatbot users' working hours.
But they found no statistically detectable average effect on recorded hours or earnings.
In the two years after ChatGPT's launch, their estimates ruled out average effects larger than about 2%.
The two studies looked at different workplaces and measured different things.
Useful to know: saving time on tasks does not guarantee a shorter working day.
Real-life examples
Real life: inside the US company.
The Berkeley Haas researchers spent eight months at a technology company of about 200 people, where staff had broad access to generative AI tools.
They observed and interviewed employees.
The researchers wrote that employees worked faster, took on a wider range of tasks, and stretched work into more hours of the day, often without being asked.
Workers increasingly kept several threads of work going at once.
Some ran several AI agents at the same time.
One of the researchers, Xingqi Maggie Ye, said this intensity can feel positive in short bursts, while the overall effect builds strain over time.
Real life: inside Danish workplaces.
The Danish researchers found that most employers in the exposed occupations had launched chatbot initiatives.
New AI-related tasks were widespread, including creating content, overseeing AI, and integrating AI into work.
According to the researchers' summary, about 8% of users in workplaces with no AI policy took on entirely new tasks, and about twice as many did so where employers actively supported AI use.
Anders Humlum told Fortune in 2025 that earlier studies focused on tasks where AI saves the most time, such as writing code. Across a broad range of occupations, the savings were much smaller.
Some chatbot users moved into higher-paying occupations where chatbots matter more, but too few to move average earnings.
The researchers' conclusion: technology can reshape work well before it shows up in earnings or recorded hours.
Real life: a US survey of AI users.
In a revised St. Louis Fed working paper (October 2025), workers who had used generative AI for work in the previous week reported saving 5.2% of their work hours on average, about 2.1 hours in a 40-hour week.
Counting all workers, including non-users, the saving was 1.4% of total work hours.
These are self-reported savings, measured differently from the Danish study.
One of the researchers, Alexander Bick, said workers' ability to use saved time to ease up is likely to be short-term, as AI use becomes more common and easier to measure.
Real life: a survey of workers and executives.
In a 2024 survey by the Upwork Research Institute, 96% of executives expected AI to raise their company's productivity.
But 77% of employees who used AI said it had added to their workload.
They pointed to more time reviewing AI-generated content, learning the tools, and being asked to do more work because of AI.
40% felt their company was asking too much of them when it came to AI.
The survey covered 2,500 respondents in the US, the UK, Australia, and Canada, including 625 full-time salaried employees. Upwork runs a platform for freelance work.
Back to Toto-Rex
So, where did Toto-Rex's saved time go? Into more work, as at the US company in the study.
The Danish data point the same way on average: time saved on tasks did not show up as fewer recorded hours.
In the Upwork survey, employees linked the extra work to reviewing AI output, learning the tools, and being asked to do more.
Taken together, these findings suggest that saving time on tasks need not shorten the working day.
Summary
In short: AI can speed up tasks without shortening the working day.
One US company saw work intensify, and a Danish study found small time savings but no detectable average change in hours or pay.
Three key terms
Work intensification
People work faster, with more demands or effort. For example: taking on more tasks in the same day.
Self-reported data
People describe their own experience. It can differ from what official records measure.
Time saved on tasks can coexist with unchanged recorded hours.
In the Danish study, the researchers used workers' reports to estimate time saved on tasks equal to about 3% of chatbot users'
working hours.
Null result
The study did not detect a statistically significant effect. That does not prove the true effect is zero.
How much a study can rule out depends on how precise its estimates are.
In the Danish study, the estimates were precise enough to rule out average effects larger than about 2% on recorded hours and earnings.
Further reading
Start with the Harvard Business Review article on the eight-month study at one US tech company.
Then the Danish study's working paper, for its full results.
For more on the US study, see the Berkeley Haas interview with the researchers.
Sources
Ranganathan and Ye, 'AI Doesn't Reduce Work – It Intensifies It', Harvard Business Review, 9 February 2026. Humlum and Vestergaard, 'Still Waters, Rapid Currents: Early Labor Market Transformation under Generative AI', NBER Working Paper 33777, revised March 2026.
Berkeley Haas Newsroom, 'AI promised to free up workers' time. UC Berkeley Haas researchers found the opposite.', 18 February 2026. Fortune, interview with Anders Humlum, May 2025.
Bick, Blandin, and Deming, 'The Rapid Adoption of Generative AI', Federal Reserve Bank of St. Louis Working Paper 2024-027F, revised 27 October 2025. Federal Reserve Bank of St. Louis, 'Generative AI, Productivity and the Future of Work', October 2025.
Upwork Research Institute, press release of 23 July 2024.