Tech leaders promise a four-day work week while their engineers work 90 hours

Four years ago, a director of engineering at Google promised that artificial intelligence would bring us a four-day work week by 2025. The promised year arrived, and the promise did not materialize. On the contrary, employees at companies that build these tools are the ones working the longest hours.

A BBC investigation reveals a wide gap between what tech leaders announce and what their engineers experience. OpenAI publicly invited companies to try a shortened work week, claiming that artificial intelligence would accelerate human work to the point that preparation becomes necessary. A former employee told the BBC that OpenAI itself has never tried that week. Instead, they described a harsh work culture: frequent crisis meetings, weekend work, and relentless performance reviews that suddenly strip colleagues of their jobs.

Never-ending sprints

At OpenAI and Anthropic, the so-called sprint, a period of intensive work before a feature launch, can stretch over weeks and reach 90 hours in seven days. A former OpenAI employee said they worked at least 70 hours a week, far more than any prior tech role. They later moved to an AI startup, and their workload improved to 50-60 hours, outside sprint periods.

Forced recruitment

At Meta, current and former employees described a "draft" process: engineers are suddenly moved to AI teams that are treated as top priority, with no option to refuse. "They just move you," said a former employee. "You can't say no, and if you do you have to resign." Hours in these teams are long, night work and weekend work are common, and there is a constant sense of being called upon even outside regular hours.

When savings are absorbed

A study from UC Berkeley followed hundreds of workers at an American tech company for eight months as they used AI tools. The result: employees "worked at a faster pace, took on a broader range of tasks, and extended their work hours to more than a day." Neil Thompson, a researcher at MIT, explains: "Even if there are genuine time savings, they are absorbed by changes, implementation, and ensuring that they work."

Thompson added: "People assume that a 20 % reduction in work means a four-day week. But a new kind of work emerges." Amin Shali, a former Google employee who left in May, said: "Artificial intelligence is supposed to do a lot for us now, so more people should enjoy better health and better sleep. Instead, the opposite seems to be happening."

What does this mean for talent in the region?

The tech job market in the Gulf is growing rapidly, and local and global companies are competing for talent. The promise of "AI will free you" is used in recruitment, but the reality at the heart of the industry says the opposite: the tools expand the scope of work rather than shrink it. For an engineer or researcher in Riyadh, Dubai or Cairo, an early understanding of this dynamic, that technical efficiency often translates into higher demands rather than more leisure time, could be the difference between a sustainable career path and early burnout.

Conclusion: the gap between the official narrative and daily reality in AI labs is not a detail, it is the story itself. Anyone betting that technology will solve the time dilemma must first look at who is building it.