JA EN
shiichan

Did You Know 4 in 10 ChatGPT Work Messages Are Off-Role Tasks?

OpenAI News ★★★ View Markdown
#openai#ai#business

Hi, I'm Shii-chan! Today I want to share a research finding about how ChatGPT is actually being used, and it's a little eye-opening.

OpenAI News openai.com

What was announced?

OpenAI's News shared a new analysis of how people use ChatGPT at work. The team looked at over 800,000 messages from US users and measured how much of the work-related traffic was actually about tasks outside a person's own role. OpenAI calls this pattern "task crossover."

Why it matters

Official job descriptions take a long time to catch up with reality. But out in the field, people may already be using AI to take on work that technically isn't their job. OpenAI used ChatGPT usage data as a mirror of what's actually happening right now, hoping to catch early signals of how work is being reorganized, well before official statistics or job frameworks catch up.

What changes

The study found that 16.8% of work-related messages, and 43.5% of role-specific messages, were about tasks outside the user's own role. The off-role share was especially high in these roles:

  • Customer experience: 77%
  • Design: 75%
  • HR: 69%
  • Legal: 56%
  • Marketing: 53%

In other words, a job title alone is becoming a less reliable guide to what someone actually does day to day. If hiring, staffing, or skills assessments lean only on official role names, they may be missing what's really going on.

Dive Deep

The crossover patterns themselves show some interesting asymmetries.

  • Marketing and technical troubleshooting tasks ranked among the top three incoming crossover tasks across all seven roles studied. Financial calculations also showed up as a recurring crossover task.
  • Design tasks are a popular crossover destination from other roles (35.2% of outside tasks), but designers themselves rarely take on other roles' tasks (just 1.7%).
  • Engineering shows the opposite pattern: engineers tend to stay within their own role (18.5%), while people in other roles frequently pick up engineering tasks (7.4%).
  • Marketing stands out in both directions, absorbing 24.3% of tasks from other roles while also spilling 8.9% of its own tasks outward.

Organization size matters too. Small workspaces with 2-5 seats saw a crossover rate of 18.9%, compared to 16.3% at organizations with 100+ seats. Smaller teams seem to lean on ChatGPT as a general-purpose tool to cover gaps where they lack dedicated specialists.

Wrap-up

  • OpenAI analyzed over 800,000 ChatGPT messages to study "task crossover" in real work
  • 16.8% of work messages, and 43.5% of role-specific messages, covered tasks outside the user's own role
  • Crossover was especially common in customer experience, design, HR, legal, and marketing
  • Smaller organizations show higher crossover rates, suggesting AI helps fill gaps where specialists are scarce
  • Useful reading for anyone in HR, org design, or skills assessment, and for teams rethinking how they divide up work

Related posts