"AI Is Said to Boost Work Efficiency"...But AI Developers Work Up to 90 Hours a Week

Both Workload and Working Hours Increase After AI Adoption
Employees Reassigned to AI Tasks...Night and Weekend Work Required

Contrary to predictions by big tech companies that artificial intelligence (AI) will boost work efficiency and reduce working hours, employees who actually develop AI are reportedly suffering from long working hours.

Photo is not directly related to specific content of the article. Pixabay

AI Is Supposed to Reduce Working Hours...Some Work Over 90 Hours a Week

The BBC reported on August 10 (local time) about the working conditions at major technology firms that have invested heavily in developing AI.

An engineering director at Google predicted four years ago that AI would allow for a four-day workweek by 2025. OpenAI also suggested this year that companies should experiment with a four-day workweek without pay cuts, claiming AI would significantly reduce human working hours.

However, the reality on the ground looks quite different.

A former OpenAI employee who left the company last year told the BBC that the company had never actually implemented a four-day workweek during their tenure. On the contrary, there were frequent emergency meetings and weekend work, with fierce performance evaluations and sudden layoffs among coworkers.

This employee said, "At OpenAI, I worked more than 70 hours per week," adding, "At the AI startup where I currently work, I usually put in about 50 to 60 hours a week, but during 'sprint' periods ahead of new product or feature launches, the working hours increase significantly."

According to sources interviewed by the BBC, these sprints at OpenAI and Anthropic have sometimes lasted for several weeks. Some employees reportedly worked more than 90 hours in a seven-day period. Under U.S. federal law, there is no limit to the number of hours employees aged 16 and older can work in a day or week.

The photo is not related to the specific content of the article. Pixabay

Workload Increases Due to AI Output Review...Saved Time Is Filled with New Work

Employees at Meta reported that as AI initiatives became urgent tasks this year, some were transferred to related teams regardless of their own preferences. In these teams, employees often worked late into the night and responded to work needs even on weekends. According to these employees, there was no end to the task of developing systems where AI could replace human work.

A research team at UC Berkeley tracked hundreds of employees at a tech company over an eight-month period and found that while those using AI worked faster, the scope of their responsibilities expanded and the total number of daily working hours also increased. In particular, the need to continually review outputs generated by AI was cited as one of the main reasons for the increased workload.

Neil Thompson, an MIT researcher, told the BBC, "Even if AI saves you time, that time is offset by the additional work of adopting, verifying, or checking for errors in these new tools, so any reduction in hours is immediately filled with more work."

This content was produced with the assistance of AI translation services.