A group of researchers who wanted to see how AI affects us at work found that it actually makes us work harder

Artificial intelligence (AI)-based services and tools, such as OpenAI's ChatGPT, have been touted as tools capable of bringing productivity gains, but their real effect has been to make people work harder and benefit less from their work, an international group of researchers says in a study cited by The Register.

In a paper entitled In “AI and the Extended Workday: Productivity, Contracting Efficiency, and the Distribution of Economic Rents,” economists Wei Jiang (Emory University), Junyoung Park (Auburn University), Rachel Xiao (Fordham University), and Shen Zhang (Seton Hall University) looked at how occupational exposure to artificial intelligence influences work time.

“When ChatGPT came out, we were all captivated by how powerful it is, how many things it can do,” said Wei Jiang, a finance professor at Emory University in an interview with The Register. “So, like other people, we imagined that if AI made part of our work, we would be able to work less. And yet I found that I was actually working more. So I checked with a few friends, and they all said, 'Hey, we work more, too,'” she said.

A study started from a researcher's curiosity

Jiang said she wanted to find out if her and her colleagues' experience reflected a larger trend, which is why she and her co-authors decided to analyze data from American Time Use Survey – an annual survey by the US Bureau of Labor Statistics that tracks how Americans spend their time.

“People fill out 15-minute surveys [despre activitățile desfășurate] the previous day,” Jiang explained. “So you can actually track how much people work, and we correlated each occupation with its degree of exposure to AI. What we found is that if your workplace is more exposed to AI, you actually work more,” she explained.

“Workers in occupations with a greater degree of exposure to generative artificial intelligence experienced a significant increase in work hours and a decrease in leisure time after the launch of ChatGPT,” states the study signed by her and her colleagues.

Based on data from 2022 and 2023 (the years when ChatGPT became widely used), a significant increase in job exposure to generative AI corresponded to 3.15 additional hours of work per week and a 3.20 hour reduction in weekly leisure time.

Artificial intelligence really increases productivity

Jiang takes it for granted that artificial intelligence increases productivity, although several recent studies suggest otherwise.

“I don't think it's much of a controversy to say that AI is bringing productivity gains,” she said. “However, the problem is that employees may not benefit from these gains. I think the misunderstanding arises as to who actually enjoys them,” the economist added.

She says that in theory three groups could benefit from increased productivity: organizations and their shareholders, employees and consumers.

But she says that in highly competitive labor markets, employees have little bargaining power to demand the productivity benefits of technology. Furthermore, where the product market is competitive, much of the economic benefit is passed on to consumers.

In other words, to the extent that AI increases productivity, the added value tends to go to employers or consumers, not workers.

“I think in general people tend to agree that AI technology brings the biggest gains to consumers and businesses, not to the broad masses of employees,” Jiang told The Register.

“These led to longer working hours”

The authors also state that extended working hours are also explained by the role of AI in supervising workers to enforce higher productivity. “Remote workers with greater exposure to AI-based surveillance technologies work longer hours post-pandemic,” the study said. He notes that the effect is not seen among the self-employed.

“The combined results suggest that while AI-driven productivity gains promise greater efficiency, they have led to longer working hours and lower employee satisfaction—especially in competitive markets and in occupations with greater AI complementarity—contradicting the traditional expectation that technology frees people from longer workdays,” the study concludes.