CoChat
Sign Up
Cited by 244AI in Service InteractionsArtificial Intelligence ApplicationsEthics and Social Impacts of AI

How People Use ChatGPT

Open access
Chatterji, Aaron, Cunningham, Thomas, Deming, David, Hitzig, Zoe, Ong, Christopher, Shan, Carl Yan, Wadman, Kevin·2025·National Bureau of Economic Research
DOI 10.3386/w34255
Open this paper in CoChat — free account

Abstract

Despite the rapid adoption of LLM chatbots, little is known about how they are used.We document the growth of ChatGPT's consumer product from its launch in November 2022 through July 2025, when it had been adopted by around 10% of the world's adult population.Early adopters were disproportionately male but the gender gap has narrowed dramatically, and we find higher growth rates in lower-income countries.Using a privacy-preserving automated pipeline, we classify usage patterns within a representative sample of ChatGPT conversations.We find steady growth in work-related messages but even faster growth in non-work-related messages, which have grown from 53% to more than 70% of all usage.Work usage is more common for educated users in highly-paid professional occupations.We classify messages by conversation topic and find that "Practical Guidance," "Seeking Information," and "Writing" are the three most common topics and collectively account for nearly 80% of all conversations.Writing dominates work-related tasks, highlighting chatbots' unique ability to generate digital outputs compared to traditional search engines.Computer programming and self-expression both represent relatively small shares of use.Overall, we find that ChatGPT provides economic value through decision support, which is especially important in knowledge-intensive jobs.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.3386/w34255

Cites

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Students predominantly use systems such as ChatGPT for information retrieval, clarification of concepts, and obtaining direct answers rather than for deeper reflective or metacognitive engagement.

    “which suggests that students predominantly use systems such as ChatGPT for information retrieval, clarification of concepts, and obtaining direct answers rather than for deeper reflective or metacognitive engagement (Chat -terji et al., 2025 ; Vodafone Stiftung Deutschland gGmbH, 2024).”

    From Enhancing School Students' Self-Regulated Learning through Generative AI Support: A Randomized Controlled Trial (2026) · cites this paper as Chatterji et al., 2025

    Section 1 reports that seeking information and practical guidance are common ChatGPT uses, and defines practical guidance to include tutoring and teaching. The paper analyzes ChatGPT users generally, not students specifically, and does not compare students’ requests for answers or concept clarification with deeper reflective or metacognitive engagement. Its measure of personal reflection is not a measure of metacognitive learning. (LLM, conf=0.83)

    Experimental — accuracy not yet measured

  • Millions of users now engage in sustained, relational interactions with AI chatbots, often describing these systems in interpersonal terms.

    “According to a recent report by OpenAI, millions of users now engage in sustained, relational interactions with AI chatbots, often describing these systems in interpersonal terms (Chatterji et al., 2025).”

    From Intimate Strangers by Design: A Uses and Gratifications Analysis of AI Companionship (2026) · cites this paper as Chatterji et al., 2025

    The Introduction reports 700 million weekly ChatGPT users and says 1.9% of messages concern relationships and personal reflection. Those figures do not establish that millions of distinct users engage in sustained relational interactions. The Appendix includes an interpersonal example (“I want you to be my AI girlfriend”), but does not show that users often describe the chatbot in such terms. (LLM, conf=0.84)

    Experimental — accuracy not yet measured

  • Conversational AI systems are increasingly used not only as task-oriented tools but also as relational technologies with which users form sustained emotional connections.

    “Conversational artificial intelligence (AI) systems are increasingly used not only as task-oriented tools, for information retrieval, coding assistance, or productivity support, but as relational technologies with which users form sustained emotional connections.”

    From Intimate Strangers by Design: A Uses and Gratifications Analysis of AI Companionship (2026) · cites this paper as Chatterji et al., 2025

    The paper’s usage analysis reports task-oriented uses such as information seeking, writing, and programming, while also identifying a small share of messages about relationships and personal reflection (Section 1; Appendix B). It does not establish that users form sustained emotional connections with ChatGPT; discussing feelings or seeking relationship advice is not evidence of such a connection. (LLM, conf=0.88)

    Experimental — accuracy not yet measured

  • Large language models (LLMs) are increasingly used as knowledge tools for answering questions, explaining concepts, drafting arguments, and supporting inquiry.

    “Large language models (LLMs) are increasingly used not only as predictive tools, but as knowledge tools for answering questions, explaining concepts, drafting arguments, and supporting inquiry (Chatterji et al., 2025).”

    From On Epistemic Diversity in Large Language Models (2026) · cites this paper as Chatterji et al., 2025

  • Information seeking is among the most common conversational-AI uses.

    “Information seeking is among the most common conversational-AI uses [7]”

    From The New Shape of Search: How Conversational AI Recomposes Information Seeking (2026) · cites this paper as [7]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Still being checked.

Research works printed but not linked: 24

  1. Acemoglu, Daron, "The Simple Macroeconomics of AI," Technical Report 32487, National Bureau of Economic Research, Cambridge, MA May 2024.
  2. Autor, David H., Frank Levy, and Richard J. Murnane, "The Skill Content of Recent Tech- nological Change: An Empirical Exploration," Quarterly Journal of Economics, November 2003, 118 (4), 1279-1333.
  3. Bengio, Yoshua, Aaron Courville, and Pascal Vincent, "Representation Learning: A Review and New Perspectives," 2014.
  4. Bick, Alexander, Adam Blandin, and David J. Deming, "The Rapid Adoption of Generative AI," Technical Report 32966, National Bureau of Economic Research, Cambridge, MA September 2024.
  5. Caplin, Andrew, David J. Deming, Søren Leth-Petersen, and Ben Weidmann, "Economic Decision-Making Skill Predicts Income in Two Countries," NBER Working Paper 31674, National Bureau of Economic Research, Cambridge, MA September 2023. Revised May 2024.
  6. Carnehl, Christoph and Johannes Schneider, "A Quest for Knowledge," Econometrica, March 2025, 93 (2), 623-659. Published March 2025.
  7. Chetty, Raj, Matthew O. Jackson, Theresa Kuchler, Johannes Stroebel, Nathaniel Hendren, Robert B. Fluegge, Sara Gong, Federico Gonzalez, Armelle Grondin, Matthew Jacob, Drew Johnston, Martin Koenen, Eduardo Laguna-Muggenburg, Flo- rian Mudekereza, Tom Rutter, Nicolaj Thor, Wilbur Townsend, Ruby Zhang, Mike Bailey, Pablo Barberá, Monica Bhole, and Nils Wernerfelt, "Social Capital I: Measurement and Associations with Economic Mobility," Nature, 2022, 608 (7923), 108-121.
  8. Chiang, Wei-Lin, Lianmin Zheng, Ying Sheng, Anastasios N. Angelopoulos, Tianle Li, Dacheng Li, Banghua Zhu, Hao Zhang, Michael I. Jordan, Joseph E. Gonzalez, and Ion Stoica, "Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference," in "Proceedings of the 41st International Conference on Machine Learning" ICML'24 JMLR.org Vienna, Austria 2024, pp. 8359-8388.
  9. Collis, Avinash and Erik Brynjolfsson, "AI's Overlooked $97 Billion Contribution to the Econ- omy," Wall Street Journal, August 2025.
  10. Deming, David J., "The Growing Importance of Decision-Making on the Job," NBER Working Paper 28733, National Bureau of Economic Research, Cambridge, MA April 2021.
  11. Eloundou, Tyna, Alex Beutel, David G. Robinson, Keren Gu, Anna-Luisa Brakman, Pamela Mishkin, Meghan Shah, Johannes Heidecke, Lilian Weng, and Adam Tau- man Kalai, "First-Person Fairness in Chatbots," in "The Thirteenth International Conference on Learning Representations" ICLR 2024 Singapore 2025.
  12. Garicano, Luis, "Hierarchies and the Organization of Knowledge in Production," Journal of Political Economy, October 2000, 108 (5), 874-904.
  13. and Esteban Rossi-Hansberg, "Organization and Inequality in a Knowledge Economy," Quar- terly Journal of Economics, November 2006, 121 (4), 1383-1435.
  14. Hartley, Jonathan, Filip Jolevski, Vitor Melo, and Brendan Moore, "The Labor Market Effects of Generative Artificial Intelligence," SSRN Working Paper, 2025. Posted: December 18, 2024; last revised: September 9, 2025.
  15. Hendrycks, Dan, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt, "Measuring Massive Multitask Language Understanding," in "Proceedings of the International Conference on Learning Representations (ICLR)" 2021.
  16. Hofstra, Bas, Vivek V. Kulkarni, Sebastian Munoz-Najar Galvez, Bryan He, Dan Juraf- sky, and Daniel A. McFarland, "The Diversity-Innovation Paradox in Science," Proceedings of the National Academy of Sciences, 2020, 117 (17), 9284-9291.
  17. and , "The Unequal Adoption of ChatGPT Exacerbates Existing Inequalities among Workers," Proceedings of the National Academy of Sciences, 2025, 122 (1), e2414972121.
  18. Ide, Enrique and Eduard Talamas, "Artificial Intelligence in the Knowledge Economy," Journal of Political Economy, June 2025, 9 (122), null.
  19. Korinek, Anton and Donghyun Suh, "Scenarios for the Transition to AI," Technical Report 32255, National Bureau of Economic Research, Cambridge, MA March 2024.
  20. Lambert, Nathan, Jacob Morrison, Valentina Pyatkin, Shengyi Huang, Hamish Ivison, Faeze Brahman, Lester James V Miranda, Alisa Liu, Nouha Dziri, Shane Lyu et al., "Tulu 3: Pushing frontiers in open language model post-training," arXiv preprint arXiv:2411.15124, 2024. · arXiv
  21. Ling, Yier and Alex Imas, "Underreporting of AI use: The role of social desirability bias," https: //ssrn.com/abstract=5232910 May 2025. Available at SSRN: https://ssrn.com/abstract= 5232910 or http://dx.doi.org/10.2139/ssrn.5232910. · DOI
  22. OpenAI, "GPT-4 Technical Report," 2023. arXiv preprint. , "GPT-4o System Card," https://cdn.openai.com/gpt-4o-system-card.pdf 2024. , "OpenAI o1 System Card," System Card / Technical Report, arXiv December 2024. Submitted 21 December 2024. , "Expanding on What We Missed with Sycophancy," Blog Post / Technical Report, OpenAI May 2025. A detailed follow-up on the GPT-4o sycophancy rollback, outlining causes and improvements. , "GPT-5 System Card," System Card / Technical Report August 2025. GPT-5 system card, OpenAI. , "Privacy Policy," https://openai.com/policies/row-privacy-policy/ 2025. last updated June 27, 2025.
  23. Vaswani, Ashish, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin, "Attention Is All You Need," in I. Guyon, U. Von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, eds., Ad- vances in Neural Information Processing Systems, Vol. 30 of 31st Conference on Neural Information Processing Systems (NIPS)
  24. West, Jevin D., Jennifer Jacquet, Molly M. King, Shelley J. Correll, and Carl T. Bergstrom, "The Role of Gender in Scholarly Authorship," PLoS ONE, 2013, 8 (7), e66212.

Cited by

Sorted by citations

Related

Authors
Chatterji, A., Cunningham, T., Deming, D., Hitzig, Z., Ong, C., Shan, C. Y., & Wadman, K.
Affiliations
  • Duke University (US)
  • Meta (United States) (US)
  • National Bureau of Economic Research (US)
  • Harvard University (US)
  • OpenAI (United States) (US)
Corresponding author
Chatterji, Aaron
ORCID
Chatterji, Aaron
Venue
National Bureau of Economic Research
Publisher
National Bureau of Economic Research
Published
2025-09
DOI
10.3386/w34255
OpenAlex
W4414192430
Cited by
233
References
0
Citations by year
2026: 201 · 2025: 31 · 2024: 1
Type
report
Language
en
Open access
Locations
  • National Bureau of Economic Research (publishedVersion)
  • SSRN Electronic Journal (acceptedVersion)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › AI in Service Interactions
  • Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
  • Social Sciences › Safety Research › Ethics and Social Impacts of AI
Keywords
ChatGPT, large language model chatbots, chatbot adoption, usage patterns, gender gap, income gap, conversation topics, decision support, conversational AI

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review