OpenAI’s recent analysis of professional ChatGPT activity shows that cross-occupation tasks are moving from one-off experiments into regular work routines for many employees. Examining more than 1.5 million work-related messages exchanged between April and July 2026, the study finds that activities typically associated with other occupations are not only being tried by workers but are increasingly revisited and folded into day-to-day workflows.
The report identifies clear behavioral differences in how workers use AI depending on whether a request sits inside or outside their usual occupational boundary. For tasks outside a worker’s normal role, prompts tend to be shorter and workers are less likely to ask the model for explanations, step-by-step walkthroughs, specific response formats, or broad advice. Instead, they more frequently supply contextual materials—examples, documents, or background—and ask the model to check or verify information. OpenAI interprets these patterns as workers “borrowing expertise”: bringing context and asking the model to apply knowledge linked to another field rather than asking the model to teach a new discipline in depth.
To measure recurrence, OpenAI tracked a cohort of roughly 6,200 workers observed consistently from April through July. In that group, the share of occupation-specific AI activity made up by previously used cross-occupation tasks rose from 13.1% in April to 25.9% in July, indicating that cross-occupation use was becoming a larger component of observed AI activity over the four-month period. A separate matched one-month follow-up analysis reinforced this pattern: workers returned to a cross-occupation task used in the prior month 23.6% of the time, versus 8.4% for comparable workers who had not used the task in the previous month. The study reports similar recurrence gaps for tasks inside a worker’s occupation and for general tasks, suggesting a broader pattern of repeated AI-assisted activity.
Not all cross-occupation tasks show the same propensity to stick. OpenAI reports substantial variation in next-month return rates by task type, which may reflect where AI naturally integrates into recurring work, as well as workplace norms, caution, or the perceived cost of errors. The highest next-month return rates were found for customer-facing and marketing-related tasks: discussing goods or services with customers returned 54% of the time, advertising or promotional writing returned 44%, and creating marketing materials returned 37%. By contrast, explaining financial information returned at roughly a 15% rate the next month. Across all cross-occupation tasks measured, the average next-month return rate was 18.5%.
These findings suggest a practical trajectory for job change that does not necessarily require changes in job titles. OpenAI frames the movement from one-off use to repeated activity as a possible route for job transformation: workers experiment with tasks outside their traditional remit, find AI helpful, and begin to perform those activities regularly as part of their role. That can broaden the mix of tasks a position entails even if formal job classifications remain static.
For organizations planning AI adoption, the report underscores that tool access alone is not sufficient. How tasks are designed, divided, and integrated into workflows matters for realizing AI’s potential. The observed shift toward recurring cross-occupation tasks implies companies may gain by reducing friction between problem identification and task execution—through clearer work design, role expectations, and supportive processes—rather than relying solely on deploying tools.
OpenAI also signals intentions to continue researching how these patterns evolve. The report references prior Work at the Frontier research on task crossover and links to an AI Jobs Transition Framework that models how AI capabilities, human roles, and demand could shape employment outcomes. The full OpenAI report, including data and methodology behind the findings, is available on the OpenAI website, where the organization invites further study of how cross-occupation tasks evolve within AI-assisted work.
As AI tools become more accessible and workers grow more comfortable integrating them into daily practice, this study offers an early empirical window into how those tools can reshape the content of jobs. The trend toward recurring cross-occupation tasks points to a subtle but significant shift: AI can enable workers to incorporate expertise from other fields into their routines, expanding what individuals do at work even without formal role changes. Organizations that attend to both tool deployment and the design of tasks and responsibilities are likely to be better positioned to capture the benefits and manage the risks of that transition.
Source: Read the original source
Leave a Reply