Tech company work hours: Employee testimonies state that the artificial intelligence development race has pushed working hours to 70 and 90 hours a week, despite promises that the technology would shorten the workweek.
Article Index
- Promises to reduce work hours versus the field reality
- Hustle culture and crisis meetings
- Forced employee transfers and project pressures at Meta
- Erosion of time gains and increased verification burdens
- Health and psychological impacts on engineering talent
- Frequently Asked Questions
Promises to reduce work hours versus the field reality
For many years, executives at companies pumping hundreds of billions of dollars annually into developing AI tools have continued to emphasize that this technology will eventually reduce the time humans spend working. Four years ago, Google’s director of engineering predicted that AI would achieve a four-day workweek by 2025. Earlier this year, OpenAI officially urged companies to start testing a four-day workweek without changing wages, claiming the technology would accelerate human work. However, employees deny that this is being implemented.
Hustle culture and crisis meetings
A former technical employee at OpenAI revealed to the BBC that the company never tested the four-day workweek as it suggested to others. Instead, they described a stressful work culture characterized by frequent crisis meetings, weekend work, and strict performance reviews leading to sudden layoffs. They explained that they worked at least 70 hours a week, while working hours during launch sprints at companies like OpenAI and Anthropic reach over 90 hours across seven days.
Forced employee transfers and project pressures at Meta
At Meta, employees reported being suddenly transferred this year to teams working urgently on AI projects without being given the option to refuse, in what they described as “forced conscription.” The tasks of these teams involve working late into the night and on weekends, with a constant feeling of being on call at any moment. These teams work on building software engineering AI tools and developing infrastructure to measure model simulation for human jobs, making the nature of the work feel endless.
Erosion of time gains and increased verification burdens
Recent research from the University of California, Berkeley involving hundreds of employees indicates that AI tools have made workloads more exhausting; employees worked at a faster pace, took on a broader scope of tasks, and extended working hours to audit AI outputs. Neil Thompson, a researcher at the Massachusetts Institute of Technology, explained that even if there are real time savings, they are absorbed by successive changes and the addition of new tasks to ensure the tools work efficiently, creating a high-pressure environment.
Health and psychological impacts on engineering talent
Former engineers who left major companies pointed out that intensive reliance on AI has created a culture of excessive pressure that negatively affected sleep quality and the general health of employees. Instead of improving quality of life, smart tools have turned into a means to increase productivity expectations and expand daily working hours unprecedentedly in modern technological history.
This development places tech company work hours at the center of the debate, emphasizing the need to separate confirmed results from expectations that still await further testing or data.
Sources: BBC via AOL
Frequently Asked Questions
Question: Did OpenAI implement the 4-day workweek internally?
Answer: According to former employees, the company did not implement this system internally, and employees worked 70 to 90 hours a week.
Question: What does the term forced conscription for employees mean at Meta?
Answer: It refers to borrowing a military description for work pressure and the rapid transfer of resources to urgent AI teams, which alone does not prove the existence of an official policy of coercion.
Question: Why didn’t AI tools reduce actual working hours?
Answer: Due to increased output verification burdens, expanded task scopes, and the redirection of saved time toward new tasks.
