Workers are increasingly using ChatGPT to perform tasks traditionally reserved for specialists in other fields, according to new OpenAI research released Monday and shared first with Axios. The findings suggest AI is fundamentally reshaping how labor gets divided within organizations—potentially before economists can even measure the shift.

The Numbers Tell the Story

After stripping out generic requests like email drafting and calendar management, researchers found approximately 44% of occupation-specific ChatGPT messages involved tasks typically associated with a different profession. The study analyzed more than 800,000 work-related prompts from U.S. business users to arrive at this figure—making it one of the largest real-world datasets on how people actually use AI in professional settings.

Who's Crossing Over Most

Customer service representatives, designers, and HR professionals showed the highest rates of cross-occupation AI usage. Roughly three-quarters of their occupation-specific prompts involved tasks more commonly handled by specialists outside their formal roles. These workers turned to ChatGPT for everything from creating marketing materials and troubleshooting software code to performing financial calculations and explaining regulatory requirements. "The boundaries between jobs are likely already becoming more flexible due to AI," OpenAI chief economist Ronnie Chatterji told Axios, summing up the research's core implication.

Small Firms Feel the Effect First

The data reveals an interesting size disparity: nearly 19% of work-related requests at the smallest businesses crossed occupational boundaries, compared to roughly 16% at larger firms. Researchers attribute this gap to workers at smaller companies using AI to fill expertise gaps when specialists aren't available on payroll—a pattern that suggests AI adoption may accelerate consolidation among leaner organizations.

What We Don't Know Yet

The authors caution their analysis cannot determine whether AI is creating genuinely new cross-occupation work or merely helping employees handle responsibilities they already technically held. The research also doesn't measure productivity gains, output quality, or downstream hiring decisions—critical variables that would determine whether this trend represents efficiency gains or simply task shuffling.

Key Takeaways

  • Nearly half of all occupation-specific ChatGPT requests involve tasks from another profession
  • Customer service, design, and HR workers show the highest crossover rates at ~75%
  • Smaller businesses see more cross-boundary AI usage than larger enterprises
  • The trend precedes any measurable changes in official labor market statistics

The Bottom Line

This research confirms what many of us suspected: AI isn't just automating existing jobs—it's dissolving job definitions themselves. Companies that adapt their hiring models to account for these fluid boundaries will have a structural advantage over those clinging to traditional org charts.