How People Are Using ChatGPT: Analysis of 1.5 Million Conversations

OpenAI and Harvard economist David Deming have released the largest study to date on ChatGPT consumer usage, analyzing 1.5 million privacy-preserving conversations to determine how the technology creates economic value. The findings indicate that AI adoption has broadened significantly beyond early adopters, narrowing gender gaps and expanding rapidly in low- and middle-income countries.

Democratization of AI Adoption

ChatGPT usage is increasingly reflecting the general adult population as early adoption gaps close. Between January 2024 and July 2025, the share of users with typically feminine names rose from 37% to 52%.

Global accessibility has also accelerated, particularly in developing economies. As of May 2025, adoption growth rates in the lowest income countries were more than four times higher than those in the highest income countries.

Primary Use Cases: Asking, Doing, and Expressing

Three-quarters of all conversations focus on seeking information, practical guidance, and writing. While writing is the most common work-related task, coding and self-expression remain niche activities. The study categorizes usage into three primary patterns:

• Asking (49% of usage): The largest and most highly rated category, where users treat ChatGPT as an advisor for information and guidance.
• Doing (40% of usage): Task-oriented interactions, including drafting text, planning, and programming. This category accounts for approximately one-third of all work-related use.
• Expressing (11% of usage): Activities involving personal reflection, exploration, and play.

Economic Value and Productivity

ChatGPT serves a dual role as both a professional productivity tool and a personal value driver. Approximately 30% of consumer usage is work-related, while 70% is non-work related. Both categories are growing over time.

The technology creates economic value primarily through decision support, improving judgment and productivity—particularly in knowledge-intensive roles. The study notes that in some instances, ChatGPT generates value that traditional economic metrics, such as GDP, fail to capture.

Study Methodology and Privacy

This research was conducted as a National Bureau of Economic Research (NBER) working paper by OpenAI’s Economic Research team and Harvard economist David Deming. The analysis utilized automated tools to categorize usage patterns across 1.5 million conversations without human review of message content to ensure user privacy.