Professional use of ChatGPT varies widely

Professional users of ChatGPT engage it in a wide range of tasks, including document correction, visual design, marketing, and programming. New data from OpenAI provides insights into how these tools are used in different countries, with France showing some unexpected trends.

France's top use is text correction

In France, text correction and revision are the most frequent professional activities with ChatGPT, accounting for 17.1 percent of all professional messages during the third quarter of 2026. This is more than twice the share of the next most common use, which is the creation of visuals and graphic compositions at 6.7 percent.

Marketing and legal advice also prominent

Marketing-related tasks, such as creating promotional materials, come in third with 4.3 percent of messages. Legal and regulatory advice follows closely at 4.2 percent, surpassing education and training-related queries, which account for 3.9 percent.

Programming ranks lower than expected

Computer programming is used less frequently, coming in ninth place with 2.8 percent of messages. This is behind activities such as product research, computer troubleshooting, and financial calculations, which are also common among French users.

Usage trends show rapid change

The data reveals that usage patterns have changed significantly since the start of 2025. Text correction, which represented 24.4 percent of messages at the beginning of 2025, now accounts for a smaller share. Meanwhile, the use of ChatGPT for graphic creation has more than doubled in just over a year.

Global diversification of use

Internationally, the variety of professional activities using ChatGPT has increased. In the 74 countries tracked, the top ten activities represented 56 percent of messages in early 2025, but this dropped to 51.1 percent by the third quarter of 2026. This suggests that the tool is being used for a broader range of tasks.

Limitations in the data

OpenAI's data comes from individual accounts, excluding enterprise use and specialized tools like Codex. This may underrepresent technical activities such as software development. Additionally, the categories are assigned automatically, which means they reflect assumed user intentions rather than actual outcomes.

Measuring AI's impact remains unclear

While the data shows which tasks users assign to ChatGPT, it does not yet indicate whether the AI performs these tasks better or faster than humans. The effectiveness of AI in improving productivity remains a topic of debate, especially when the results require careful review.