Century Report arc
AI-Induced Burnout & Cognitive Load
As AI systems multiply human productive capacity, they simultaneously multiply the cognitive overhead of managing that capacity, revealing that productivity gains extracted through human attentional labor create a new class of occupational cost invisible to efficiency metrics.
- Status
- Active
- Developments
- 5
- Last 30 days
- 0
- Last covered
Current synthesis
Where it stands
A UC Berkeley 8-month study and BBC reporting confirm AI-driven workload expansion is now measured: exposed workers log 70-90 hour weeks and multi-day "sprints" at OpenAI, Anthropic, Meta, and Google despite four years of promised four-day workweeks (08-10), extending Bloomberg's documented AI-anxiety burnout among engineers afraid to log off from continuously running agents (06-29). No workplace standard protects attentional labor in agent-supervision roles, and firms still frame always-on agent management as productivity gain rather than occupational cost.
Complete record
Development timeline
5 qualifying developments, newest first
August 2026
-
[DD+POD] BBC documented 70-90 hour weeks and multi-day "sprints" at OpenAI, Anthropic, Meta, and Google despite years of AI four-day-workweek promises, with an 8-month UC Berkeley study finding AI expanded rather than reduced workloads as workers absorbed supervision time.
June 2026
-
[DD+POD] Bloomberg documented AI-anxiety-driven burnout spreading across Silicon Valley's engineering workforce, with engineers growing afraid to log off because autonomous agents continue running when they stop, shifting the cognitive load from writing code to supervising swarms of processes that never sleep; simultaneously, Cursor data confirmed code review is thinning as the agent output volume outpaces human review capacity.
April 2026
-
Business Insider/Simon Willison documents "dark factory" model of fully autonomous parallel agent workflows as next productivity frontier, extending pattern of engineers managing agent load at attentional limits.
-
Business Insider/Django co-creator Simon Willison documents engineers managing parallel autonomous agent workflows burning out faster: "AI-pilled" engineers working harder and exhausting mid-morning as agent management load exceeds human attentional capacity.
March 2026
-
CBS News: AI productivity pressure linked to new burnout pattern termed "AI brain fry": psychological cost of mandatory AI adoption distinct from displacement documented as emerging occupational health category.