ChatGPT was launched three years ago and its numbers are remarkable: by July 2025, this artificial intelligence chatbot had been used weekly by more than 700 million users, who collectively sent over 2.5 billion messages per day, which amounts to approximately 29,000 messages per second.
Despite these statistics, there was still no comprehensive understanding of how users were utilizing this AI. For this reason, the Economic Research Team at OpenAI and Harvard economist David Deming joined forces to analyze 1.5 million consumer conversations since ChatGPT became available to the public.
Among the key findings, we highlight:
The study, based on 1.5 million conversations and referencing the 700 million weekly active users as of July 2025, demonstrates that adoption has become strongly democratized. Gender gaps have been drastically reduced: in January 2024, only 37% of users had typically feminine names, whereas by July 2025, the figure had already exceeded 52%.
Furthermore, usage is global and is expanding especially rapidly in low- and middle-income countries, where adoption rates are more than four times higher than in high-income countries.
Age is also a differentiating factor: nearly half of adult messages come from users under the age of 26, reflecting a greater affinity among younger individuals for this technology.
Seventy percent of queries are not work-related, even though both categories (work-related and non-work-related) are experiencing growth. Three quarters of conversations revolve around practical guidance, writing, and information searching, which together represent 78% of messages.
Writing is the main work-related use, accounting for 42% of messages related to employment and over half among management and business users; of these, two thirds focus on modifying existing texts rather than creating new ones from scratch.
When classified by type of interaction, “Asking” messages stand out (49%), where users seek guidance, advice, or information, followed by “Doing” (40%), which includes tasks such as planning, writing, and programming, and finally “Expressing” (11%), encompassing personal reflection and exploration.
Finally, “Asking” messages have grown more rapidly than “Doing” messages over the past year and receive better quality ratings, both through direct user feedback and satisfaction models.
Since its launch in November 2022, ChatGPT has come to generate more than 2.5 billion daily messages (approximately 29,000 per second). Its usage has evolved in parallel with improvements to the model and the expansion of use cases, with increasing diversification between professional (30%) and personal (70%) applications.
The most accelerated growth can be observed in non-work-related queries, which points to personal and well-being benefits that are difficult to capture in traditional economic metrics such as GDP.
In the professional sphere, ChatGPT is establishing itself as an advisor and research assistant rather than a simple task executor, providing support for decision-making and enhancing productivity in knowledge-intensive jobs.
According to OpenAI estimates, users in the United States would need to be compensated $98 to forego generative AI for one month, implying a surplus of at least $97 billion per year. This figure reflects the tremendous economic and social value that ChatGPT has generated in only three years of existence.
Photo: Canva
Your email address will not be published. Required fields are marked *
Δ