- Ethnographic research at UC Berkeley Haas shows that generative AI significantly boosts productivity in “micro moments” such as prompting and experimentation, but makes workers feel busier and harder to disconnect.
- Many tech engineers report working longer hours despite completing tasks faster, leading to a markedly increased risk of burnout.
- Economic theory points to two effects: the “income effect” which encourages working fewer hours, and the “substitution effect” which drives working more because each hour generates greater value.
- As AI spreads, the value of skills may decrease rapidly—similar to the “turning lead into gold” analogy—forcing workers to speed up before the market becomes saturated.
- Businesses may cut staff and retain only a few AI managers, creating a winner-take-all competitive environment.
- Technological history, such as email or PowerPoint, shows local productivity gains but also the creation of more “busywork” and job pressure.
- AI makes workers more multi-tasking, increasing cognitive load and the feeling of constantly having to process multiple workflows simultaneously.
- Workers gradually take on tasks that previously belonged to others, leading to an uncontrolled expansion of job scope.
- AI also causes break times to be invaded as small, fragmented tasks creep into short rest periods.
- Proposed solutions include designing work and systems to be less like “greedy jobs,” while controlling AI usage in a planned rather than constant manner.
📌 Generative AI is not just a productivity tool but also changes the structure of work: increasing speed while lengthening working hours, expanding responsibilities, and increasing cognitive pressure. From UC Berkeley research to historical examples like email, the general trend is that as productivity rises, the feeling of overload increases accordingly. Without proper job design and AI usage discipline, a 10x gain in speed could be traded for a significantly higher risk of burnout.
