Researchers looked at whether messages appear to be work-related or non-work-related. While both types of use have increased, non-work messages account for over 70% of messages with ChatGPT.
Most messages fall into three categories:
- 28.1%: Practical guidance (e.g., customized guidance for a user, such as tutoring, teaching, or how-to advice)
- 21.3%: Seeking information (e.g., factual information about people, current events, products, or recipes)
- 28.3%: Writing (e.g., automated generation of communication and editing, critiquing user-provided text)
Other uses include:
- 7.5%: Technical help (e.g., calculation, computer programming, data analysis)
- 6%: Multimedia (e.g., create an image or other media)
- 4.3%: Self-expression (e.g., relationships, personal reflection, greetings, games, or role play)
- 4.6%: Other/Unknown
While there is increasing awareness about how GenAI is being used by some for emotional support and companionship, these topics appear to account for a small proportion of the overall use cases; however, with an estimated 18 billion weekly conversations, this still represents 342 million conversations a week (1.9%) that are related to relationships or personal reflection. It is important to keep in mind that general-purpose large language models, like those used by ChatGPT, are not built for or intended to provide emotional support, even though a subset of the public uses them in this way. APA has publicly called for appropriate AI guardrails to safeguard the public, particularly vulnerable groups like youth. Learn more about APA’s recent testimony before Congress about the need for AI guardrails.
Results revealed that people with more advanced education and those in higher-paying jobs were more likely to use ChatGPT for work. The top use case is writing (42%), and of those queries, two-thirds of messages are for assistance in modifying text rather than generating text from scratch. Employees also sought assistance in gathering and interpreting information and with decision support.
Researchers offered a framework that sorts most user queries into the themes of Asking, Doing, and Expressing. This framing highlights how people use ChatGPT as an advisor, task assistant, and as a space for personal reflection and play.
As the public continues to adopt GenAI, it is continuing to shape many aspects of daily life—from work, learning, leisure, to accessing information. To learn more about this study, read the full reportopens in new window.