OpenAI finds ChatGPT shifting tasks across job roles

OpenAI has published research on how artificial intelligence is shifting tasks across job roles. It found that many work-related uses of ChatGPT fall outside a user's stated occupation.

The study analysed more than 800,000 messages from ChatGPT users in the United States and focused on what OpenAI called "task crossover", where people use AI for work that has traditionally belonged to another profession. It found that 16.8% of work-related messages and 43.5% of occupation-specific messages involved tasks associated with another occupation.

The findings suggest a shift in how work is divided inside organisations. Rather than passing a problem to a specialist team, workers are increasingly using AI to handle adjacent tasks themselves.

A small business owner, for example, may use AI to draft copy, review a contract or carry out basic financial analysis. Sales staff may use it to examine customer data, while marketers may turn to it for website troubleshooting that might once have gone to a developer.

Cross-role shift

The pattern was most visible after excluding generic work such as writing, summarising and scheduling, which appears across many jobs. Once those tasks were removed, customer experience workers had the highest share of outside-occupation tasks, at 77% of occupation-specific messages. Designers followed at 75%, human resources workers at 69%, legal workers at 56%, marketers at 53%, and both sales and finance workers at 40%.

Engineering workers showed the lowest share in the group studied, at 28%. Even so, engineering tasks spread widely into other roles, making technical work one of the most common forms of crossover.

Marketing and engineering tasks were the most portable across occupations. Financial calculation and technology troubleshooting appeared among the three most common outside tasks in all seven other occupation groups covered by the analysis.

Marketing work was also widely distributed. Creating marketing materials appeared across five other occupational groups and was especially common among design users.

The report drew a distinction between occupations that import tasks and those that export them. Designers were presented as a role that draws heavily on other occupations: 35.2% of designer messages involved work usually linked to another field, while design work itself accounted for only 1.7% of messages from workers in other occupations.

Engineering showed the reverse tendency. Only 18.5% of engineering messages involved tasks from other fields, but engineering tasks made up 7.4% of messages from workers in other occupations, suggesting that technical problem-solving is increasingly being taken on outside engineering teams.

Marketing stood out in both directions. Marketers devoted 24.3% of their messages to tasks from other occupations, while marketing tasks made up 8.9% of messages from workers in other fields, the highest outward share in the sample.

Business size

The study also linked task crossover to company size. Among average users, the share of outside-occupation tasks fell from 18.9% in workspaces with two to five seats to 16.3% in workspaces with more than 100 seats.

That trend did not hold among the heaviest users of the system. One possible reason, OpenAI said, is that moderate users in smaller organisations may rely on AI when they encounter work that would otherwise require another function, while heavier users may have settled into more stable workflows centred on their main role.

The report argues that smaller businesses may see stronger effects because they often have fewer specialist teams and less formal delegation. In those settings, the person who first encounters a problem may be more likely to use AI to address it directly.

Early indicator

The research is part of OpenAI Economic Research's new Work at the Frontier series, which examines how AI is altering work patterns in real time. Usage data may offer an early view of occupational change before employers alter formal job descriptions or titles.

The framework used in the report separates generic work from occupation-linked work to identify when users move into tasks beyond their traditional remit. OpenAI argued that this offers a more useful picture of workplace change than a fixed list of tasks assigned to each occupation.

That approach reflects a broader debate over whether AI mainly automates existing duties or changes the boundaries between roles. In this case, the evidence suggests the technology is enabling workers to take on tasks that would previously have required a handoff to another team.

OpenAI said the data offered "an early window into how AI may be reshaping the task content of jobs before those changes appear in job descriptions or titles."