OpenAI Research: 43.5% of Occupation-Specific ChatGPT Messages Cross Into Other Jobs
An analysis of 800,000-plus messages finds AI is quietly expanding what workers do — a pattern OpenAI calls task crossover, strongest in small businesses and design roles.
In Brief
- OpenAI’s analysis of more than 800,000 messages from U.S. ChatGPT users finds 16.8% of work-related messages involve tasks associated with a different occupation
- Among occupation-specific messages, 43.5% cross job boundaries — a pattern OpenAI calls “task crossover” and reads as an early signal of how AI is reshaping jobs
- The effect is strongest in small businesses and roles like customer experience (77%), design (75%) and HR (69%) once generic tasks are excluded
OpenAI published new research on Monday suggesting AI is quietly expanding what individual workers do. “In an analysis of more than 800,000 messages from U.S. ChatGPT users, our new research suggests that 16.8% of work-related messages and 43.5% of occupation-specific messages are about tasks associated with another occupation,” the company wrote in its report, titled “Work at the Frontier: How AI is Expanding What People Do at Work.”
OpenAI names the pattern “task crossover”: a marketer generating code, an engineer drafting legal language, a designer writing marketing copy. Tasks were classified against O*NET, the U.S. Department of Labor’s occupational database, with generic activities like writing, summarizing and scheduling excluded from the occupation-specific analysis, as The Decoder notes in its summary of the findings.
The company frames the data as a leading indicator: “offering an early window into how AI may be reshaping the task content of jobs before those changes appear in job descriptions or titles,” per the report.
Task crossover is strongest where teams are smallest
The numbers vary sharply by role. After excluding generic tasks, 77% of customer experience workers’ messages involved work associated with other occupations, followed by designers at 75%, HR professionals at 69%, legal workers at 56% and marketers at 53%, according to OpenAI’s breakdown.
Direction matters too. Design shows the starkest asymmetry: 35.2% of designers’ messages involve other occupations’ work, while design tasks appear in only 1.7% of other workers’ messages — suggesting designers use AI to reach outward far more than outsiders use it to do design. Marketing flows the other way: marketing tasks show up in 8.9% of non-marketers’ messages, the highest inward share, while engineering tasks appear in 7.4% of other workers’ conversations.
Company size shapes the pattern as well. Workspaces with two to five seats show 18.9% crossover versus 16.3% for those with 100 or more seats — consistent with the reality that small-business employees have always worn multiple hats, and AI lowers the cost of each additional hat.
An early window into AI’s real labor-market effect
The research cuts against the dominant framing of AI as a pure job destroyer. Rather than measuring displaced tasks, OpenAI’s data captures workers annexing skills adjacent to their roles — the solo founder doing passable legal review, the HR manager building internal tools. It’s a self-interested framing from a company facing scrutiny over AI’s labor impact, but the underlying usage data is among the largest samples published on how people actually deploy chatbots at work.
It also complements OpenAI’s own findings from earlier research on ChatGPT usage patterns, and lands amid intense debate over AI and employment — from white-collar layoff anxiety to studies of productivity gains. The report stops short of claiming AI creates or destroys jobs; it documents boundaries blurring.
Whether task crossover shows up in wages, titles or headcounts remains the open question. Job descriptions update slowly; usage logs update in real time. For now, OpenAI’s data suggests the first labor-market effect of AI isn’t replacement — it’s sprawl.
FAQ
What is task crossover according to OpenAI?
The pattern of workers using AI for tasks associated with occupations other than their own — like a marketer writing code — which OpenAI measured in 16.8% of work-related ChatGPT messages.
Which workers show the most task crossover?
After excluding generic tasks, customer experience roles lead at 77%, followed by designers at 75%, HR at 69%, legal at 56% and marketing at 53%.
Does the research show AI eliminating jobs?
No. The study documents workers expanding into adjacent skills rather than being displaced, though OpenAI notes usage patterns may foreshadow changes before they appear in job titles or descriptions.
