Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

OpenAI research finds cross-occupation AI tasks are becoming routine work

OpenAI's economic research team published a new study on how workers use AI beyond their traditional job descriptions and which of those activities stick. It finds that tasks crossing occupational boundaries are becoming a recurring part of how people work with AI, and argues that work design matters as much as tool access.

Published

OpenAI's economic research team published a new study on September 16 examining how workers use AI on the job, and how that use turns from one-off experimentation into a durable habit. It follows the team's earlier "Work at the Frontier" report, which documented task crossover: workers using AI for activities historically associated with another occupation.

The new question is what happens after that first crossover. Do workers come back to tasks outside their occupational boundaries, and do those activities take up a growing share of their AI use over time?

The study covered more than 1.5 million work-related ChatGPT messages from April through July 2026, tracking roughly 6,200 workers consistently across the four-month window. Cross-occupation tasks rose from 13.1% of occupation-specific AI activity in April to 25.9% in July.

Workers also prompt differently depending on whether a task sits inside or outside their role. For out-of-role tasks they write shorter prompts on average and ask for explanations, how-to guidance or specific formats less often, but they are more likely to supply examples and background information and to ask AI to check or verify something.

One reading of that pattern is that workers are borrowing expertise rather than learning a new field from scratch: they bring a concrete problem and relevant context, then ask AI to apply knowledge from another domain.

For companies, the report argues that work design deserves a place alongside access to AI tools in any AI strategy. If those cross-occupation tasks keep recurring, jobs can broaden even while their titles stay the same, which raises new questions about role definitions, performance review, training and internal collaboration.

The data covers ChatGPT usage only, so it cannot say whether these tasks were done well or created measurable business value. The next test is whether organizations redraw responsibilities around this behavior, or simply treat AI as another tool bolted onto existing roles.

Why it matters

The report shifts the conversation from tool access to work design: if cross-occupation tasks keep compounding, the changes companies need to make are in role boundaries and collaboration, not just procurement.

OpenAIEnterprise AI
Back to realtime news

Nearby Updates

All

09/16, 17:00

Cohere and Open Text to pitch Canadian-made AI agent tools to enterprises

Cohere and Open Text plan to pitch Canadian-made AI agent tools to enterprise customers, according to The Globe and Mail. The report frames the two vendors as bringing domestically built agent tooling to corporate buyers, though product specifics have not been disclosed.

09/16, 17:06

Apodex releases the Apodex v1.1 Deep Discover model

Apodex has released its Apodex v1.1 Deep Discover model, according to TestingCatalog AI News. The release is clearly identified by product name and version number, but available information does not cover parameter scale, evaluation results, or how the model is made available.

09/16, 16:50

Qwen tops the ranking of the world’s most popular open-source models

Qwen has taken the top spot among the world’s most popular open-source models, according to a report carried by Chinese outlet Drivehome. For a Chinese model family that built its position on open weights, the result is a direct read on how much the developer community is actually adopting it.

09/16, 16:49

Zhipu lifts year-end ARR guidance to $3 billion, cloud revenue share to be recognised from October

Zhipu told an analyst briefing that it is raising its year-end ARR guidance by 25% to $3 billion, according to Caixin-syndicated outlet Wallstreetcn. The company also said revenue-sharing income from cloud vendors will start being recognised in October, giving the market a concrete date for when cloud-channel revenue lands.