Updated
Updated · OpenAI · Jul 27
OpenAI Finds 43.5% of U.S. ChatGPT Work Tasks Cross Occupational Lines
Updated
Updated · OpenAI · Jul 27

OpenAI Finds 43.5% of U.S. ChatGPT Work Tasks Cross Occupational Lines

3 articles · Updated · OpenAI · Jul 27

Summary

  • OpenAI’s analysis of more than 800,000 messages found 43.5% of occupation-specific work prompts from U.S. ChatGPT users involved tasks tied to another occupation; across all work-related messages, the share was 16.8%.
  • The study labels the pattern “task crossover,” arguing AI is shifting who performs work by letting employees handle tasks that once required handoffs to specialists.
  • 77% of customer-experience messages, 75% of designers’, 69% of HR workers’, 56% of legal workers’ and 53% of marketers’ occupation-specific prompts fell outside their own roles.
  • Marketing and engineering tasks spread most widely across occupations, while financial calculation and technology troubleshooting ranked among the three most common outside tasks in all seven other occupation groups analyzed.
  • Small organizations showed more crossover among average users: outside-occupation task share was 18.9% in 2–5-seat workspaces versus 16.3% in workspaces with more than 100 seats, suggesting AI may be most useful as a generalist tool where specialist support is scarce.

Insights

If AI turns every worker into a cross-functional generalist, will deep human expertise become obsolete or our most valuable asset?
Are entry-level jobs disappearing because experienced workers are using AI to handle the basic tasks they used to delegate?
How will companies compensate employees who use AI to secretly absorb the responsibilities of three different specialized roles?