OpenAI found that 43.5% of job-specific ChatGPT messages involved tasks historically associated with another occupation.

Key Takeaways

• 43.5% of job-specific ChatGPT messages involved tasks traditionally associated with another occupation.
• Cross-occupation work accounted for most job-specific messages in five of the eight fields studied, including customer experience, human resources, legal, and marketing.
• Customer experience workers had the highest crossover rate at 77%, followed by designers at 75% and human resources workers at 69%.
• Financial calculations and computer troubleshooting appeared widely among workers outside finance and engineering.
• Among typical-volume users, cross-occupation messages accounted for 18.9% of work-related messages in workspaces with two to five seats, compared with 16.3% in workspaces with more than 100 seats.

OpenAI found that people using ChatGPT at work frequently ask it for help with tasks traditionally handled by workers in other occupations, suggesting AI is broadening what employees do across roles. The report says AI may make specialized work easier for people outside that field to attempt, but specialists may remain critical for expert judgment and review.

On July 27, OpenAI launched its new Work at the Frontier research series with a report examining this shift. The company calls it “task crossover,” meaning people use AI for work historically associated with another occupation.

The report analyzed more than 800,000 work-related messages from U.S. ChatGPT users in customer experience, design, engineering, finance, human resources, legal, marketing and sales.

OpenAI matched users to occupations using role information they had provided through ChatGPT Business. It then analyzed work-related messages from those users’ individual ChatGPT accounts. The company said ChatGPT classified the messages anonymously and researchers did not read them.

Nearly half of job-specific messages cross roles

OpenAI classified 61.5% of the messages as generic work shared across many occupations, such as writing emails or scheduling meetings.

Separately, 21.8% involved tasks associated with the user’s occupation, while 16.8% involved work associated with another occupation.

After excluding the 61.5% of “generic work” messages, cross-occupation tasks accounted for 43.5% of the remaining messages.

Cross-occupation work leads in five job groups

Cross-occupation tasks accounted for most occupation-specific messages in five of the eight groups studied.

Among customer experience workers, 77% of job-specific messages involved tasks associated with another occupation. The rate was 75% for designers and 69% for human resources workers. The rate was also above half for legal workers at 56% and marketers at 53%.

Some types of work crossed occupational boundaries more consistently than others. Calculating financial data ranked among the three most common finance-related tasks for every non-finance group. Troubleshooting computer applications or systems held the same position for every non-engineering group.

The report also found that marketing and engineering tasks appeared frequently in messages from people working in other fields. This included requests to create promotional materials, explore customer data, and troubleshoot software.

Smaller workspaces show more crossover

Typical-volume users in smaller workspaces were somewhat more likely to use ChatGPT for work associated with another occupation.

Cross-occupation tasks accounted for 18.9% of work-related messages from these users in workspaces with two to five seats. The share fell to 16.3% in workspaces with more than 100 seats.

OpenAI said workers in smaller organizations may use AI when specialist help is less readily available. That relationship between workspace size and cross-occupation use was not consistent among the most active ChatGPT users. OpenAI also cautioned that the number of ChatGPT seats in a workspace may not reflect the company’s total number of employees.

OpenAI said the Work at the Frontier series will regularly examine how AI is changing work. The report leaves open whether repeated crossover will produce lasting changes in job responsibilities. If it does, OpenAI said workers may need training to evaluate AI-assisted work outside their established expertise, while organizations may need clear processes for review and accountability.