Artificial intelligence companies have repeatedly argued that productivity gains from AI could eventually reduce the amount of time people spend working. Yet employees developing and deploying the technology at some of the industry’s largest companies have described longer hours, extended product sprints and increasing pressure to accomplish more.
The contrast comes as OpenAI, Google, Anthropic and Meta promote AI systems capable of completing increasingly complex work with less human involvement. Research into workplace AI use also suggests that productivity gains do not automatically translate into shorter working hours.
Four-Day Workweek Predictions Have Yet to Materialize
In 2022, Google Cloud senior engineering director Kamelia Aryafar predicted that AI could become the primary driver of a four-day workweek by 2025. She argued that automation would reduce repetitive tasks and allow employees to focus their time elsewhere.
OpenAI has similarly called for businesses to experiment with four-day workweeks without reducing employees’ salaries as AI becomes more capable. However, a former OpenAI technical employee told the BBC that the company itself had not tested such a schedule during their time there.
Instead, the former employee described working at least 70 hours a week, including weekends, alongside frequent crisis meetings and demanding performance reviews. Workers interviewed by the BBC said intensive development periods at OpenAI and Anthropic could sometimes reach more than 90 hours in a seven-day period.
Anthropic has meanwhile demonstrated increasingly autonomous AI systems. Its research shows that Claude Code can work for progressively longer periods without human intervention, while newer development environments are designed to let the coding agent operate unattended.
AI Teams Face Longer Development Sprints
Workers at Meta told the BBC that some employees were reassigned to urgent AI projects in a process they referred to as being “drafted.” The employees described late nights and weekend work as Meta increased investment in AI models, infrastructure and developer tools.
Similar pressure has been reported at Google. Former employee Amin Shali told the BBC that he left the company in May after internal engineering problems increasingly required him to work late hours, which he associated partly with infrastructure being redirected toward AI projects.
The accounts highlight a tension between what AI may eventually automate and the work required to build, deploy and supervise the technology today.
Research Finds AI Can Intensify Work
An eight-month study of hundreds of technology workers found that generative AI increased the amount and pace of work rather than simply saving employees time. Researchers Aruna Ranganathan and Xingqi Maggie Ye found that employees completed a broader range of tasks, worked faster and extended work further into their days.
AI-generated work also required human review, creating additional responsibilities even when the technology accelerated the initial task. Harvard Business Review has separately warned that AI can create an “always-on” environment where faster production leads to faster expectations from managers.
OpenAI’s own research points to a related effect. An analysis of more than 800,000 work-related ChatGPT messages found that AI allows employees to take on tasks that would traditionally belong to other occupations, potentially expanding the scope of what an individual worker is expected to handle.
MIT innovation scholar Neil Thompson told the BBC that productivity gains do not necessarily result in equivalent reductions in working time because organizations often fill the available capacity with new tasks. For technology companies racing to develop AI, that dynamic means greater automation can coexist with increasingly demanding workloads.
Featured image credits: Magnific.com
For more stories like it, click the +Follow button at the top of this page to follow us.