In a landmark study released by the National Bureau of Economic Research (NBER), researchers have, for the first time, provided a comprehensive look into how people are actually using ChatGPT.
The working paper, titled “How People Use ChatGPT,” draws from anonymized data gathered between May 2024 and June 2025 and offers a uniquely data-rich perspective on this transformative AI technology.
ChatGPT: A Global Phenomenon
Since its launch in November 2022, ChatGPT has become one of the fastest-growing consumer technologies in history. By mid-2025, the chatbot had over 700 million users, representing about 10% of the global adult population. At its peak, ChatGPT was processing more than 18 billion messages weekly.
Where the Conversations Are Happening
The study’s key revelation is the distribution of work versus non-work usage:
- In June 2024, about 47% of messages were work-related.
- By June 2025, only 27% were work-related, as non-work use grew faster and now comprises over 70% of all usage.
This shift suggests that ChatGPT is not only becoming a work assistant but increasingly a part of daily personal life for users.
Top Use Cases: Writing, Guidance, and Information Seeking
The researchers used machine learning classifiers to categorize messages into high-level use cases. Nearly 80% of all ChatGPT interactions fall into just three categories:
- Practical Guidance: Includes tutoring, how-to advice, and brainstorming. Highly personalized and dynamic.
- Seeking Information: Similar to traditional search, these messages are requests for factual or topical knowledge.
- Writing: Involves generating, editing, translating, and summarizing text. Dominates work-related use and represents a key differentiator from search engines.
Notably, about two-thirds of writing requests involve modifying user-submitted text rather than generating new content from scratch.
Demographics and Usage Patterns
- Professional Users: Work-related usage is more prevalent among highly educated users in high-income professional occupations.
- Gender Shift: Early adopters were predominantly male, but this gender gap has since narrowed.
- Emerging Markets: Growth is especially strong in lower-income countries, showcasing the technology’s accessibility and appeal beyond traditional tech hubs.
Smaller Than Expected: Coding and Companionship
Despite common assumptions, only 4.2% of all ChatGPT messages relate to computer programming. Social or emotional use is even smaller, with just 1.9% focused on personal reflection or relationship advice.
What It All Means
This study highlights ChatGPT’s growing role in decision-making, content creation, and personal productivity. While much attention has been paid to AI’s impact on jobs and automation, the data here suggests a more nuanced reality: people are using AI to enhance both work and daily life, often in very personal and context-specific ways.
As large language models continue to evolve, understanding how people actually use them will be crucial to designing systems that align with human needs, values, and goals. This NBER report marks a critical step in that direction.
Read the full study here: NBER Working Paper No. 34255