The Shift from Experimentation to Routine

OpenAI’s research, based on 1.5 million ChatGPT messages from April to July 2026, reveals that AI-driven 'task crossover'—where workers use AI for activities outside their traditional occupational boundaries—is evolving into a permanent feature of work. While initial AI use often involves experimentation, the data shows that workers are increasingly incorporating these cross-occupation tasks into their regular routines. Among a cohort of 6,200 workers, the proportion of cross-occupation AI activity grew from 13.1% in April to 25.9% by July.

How AI Changes Prompting Behavior

Workers interact with AI differently depending on whether the task falls within or outside their primary occupation. When operating outside their core role, workers tend to write shorter prompts and are less likely to ask for instructional guidance or formatting advice. Instead, they act as 'expertise borrowers,' providing more context, documentation, or background information to the AI to help them apply knowledge from other fields to their current problem. This suggests that AI is being used as a bridge to execute tasks that would otherwise require deep, domain-specific training.

Variability in Task Stickiness

Not all cross-occupation tasks are equally likely to become recurring responsibilities. The study found an average next-month return rate of 18.5% for cross-occupation tasks, but this varies significantly by activity:

  • High-recurrence tasks: Discussing goods/services with customers (54%), advertising/promotional writing (44%), and creating marketing materials (37%).
  • Low-recurrence tasks: Explaining financial information (15%).

These differences likely reflect the natural fit of AI within specific workflows, as well as varying workplace norms and the perceived risks associated with errors in different domains.

Implications for Job Design

This trend points toward a 'job expansion' model where the scope of a role broadens through AI-assisted task integration, even while the formal job title remains static. For organizations, this means that AI strategy must move beyond simple tool access to include intentional 'work design.' Leaders should recognize that AI is not just automating existing tasks but is actively reshaping the division of labor by allowing workers to fluidly move across traditional departmental silos.