Independent briefing Work & economy
How AI is expanding what people do at work
Usage evidence suggests AI is moving tasks across occupational boundaries, not merely automating a fixed job description.
Independent briefing of OpenAI (2026). Read the original for its full evidence, context, and wording.
Start here
Three questions this article answers
Does AI only help people perform tasks already inside their occupation?
No. A meaningful share of work-related use involves tasks normally associated with a different occupation.
What does “task crossover” mean?
It means a person can use AI to perform work—such as analysis, writing, or coding—that traditionally belonged to another specialist role.
What changes first: jobs or task boundaries?
The evidence points to task boundaries changing first, as people assemble a broader mix of capabilities inside existing roles.
Go deeper
Three related questions it doesn’t answer—matched to established papers
Which occupations are most exposed to language-model capabilities?
A task-level exposure framework maps model capabilities onto occupations.
Paper answering this questionGPTs are GPTsDo cross-boundary tasks produce reliable business value?
This landmark field experiment shows large gains inside the model frontier and worse performance outside it.
Paper answering this questionNavigating the Jagged Technological FrontierCould AI reduce or deepen inequality between workers?
The study measures both average productivity and the distribution of gains.
Paper answering this questionExperimental Evidence on the Productivity Effects of Generative Artificial IntelligenceThe useful summary
What to take away
The article examines a large sample of work-related interactions and asks what occupation normally owns the task being performed. Its most useful finding is that AI use frequently crosses those boundaries. People are not simply doing their old job faster; they are reaching into adjacent forms of work.
That matters because job titles are slow-moving containers while tasks are fluid. AI may first reshape roles by allowing a marketer to analyze data, a founder to prototype software, or an analyst to produce clearer communication. But crossover is not the same as mastery: review, domain knowledge, and awareness of the model’s uneven capability remain essential.
Final conclusion
The three answers, brought together
AI is already helping people work beyond formal job boundaries. Task crossover describes that shift, and it suggests that roles may broaden before occupations disappear. The opportunity is greater access to capability; the risk is confusing access with verified expertise.