Stocks● Neutral
OpenAI Study: 43% of Work Prompts Cross Job Lines
New OpenAI research finds 43.5% of occupation-specific ChatGPT requests involve tasks from other professions. As tenure grows, crossover compounds, forcing companies to redraw job descriptions.
Continue with
For decades, the org chart worked like a map of who knew what. Marketing didn't touch contracts. Customer service didn't run financial calculations. Those boundaries held because expertise lived in specific people, and getting specialized work done meant calling in the right one. New research from OpenAI suggests generative AI is dissolving those lines fast enough that job titles are starting to lag behind what people actually do all day.
The evidence comes straight from OpenAI. The company examined more than 800,000 work-related ChatGPT messages from U.S. users and found that 43.5% of occupation-specific requests involved tasks traditionally associated with a different profession, according to the company's report, the first installment in a new "Work at the Frontier" series published July 27. OpenAI has a name for the phenomenon: task crossover.
Crossover reached 77% among customer experience workers, the highest of any role in the study, followed by 75% among designers, 69% in human resources, 56% in legal and 53% in marketing, OpenAI's research found. In practice, that's a customer experience worker running a financial calculation once handled by an analyst, or an HR professional troubleshooting a problem once sent to IT.
Direction matters as much as size, and design shows the pattern most starkly. Even counting every message they send, not just occupation-specific ones, designers pulled outside work in 35.2% of their prompts. Design expertise almost never travels the other way: just 1.7% of workers in other fields touched on design at all, OpenAI found.
Engineering runs the opposite direction: only 18.5% of engineering messages involve outside fields, the tightest boundary of any role OpenAI measured. But engineering-style tasks such as troubleshooting account for 7.4% of messages among non-engineers, suggesting more employees are patching technical problems themselves rather than filing a help desk ticket.
Marketing lands in the middle, and it's the most porous role in either direction. Marketers put 24.3% of their messages toward other occupations' work, while marketing tasks bled into 8.9% of messages from other fields, the highest outward share OpenAI recorded.