IDEAS home Printed from https://ideas.repec.org/p/arx/papers/2312.15524.html
   My bibliography  Save this paper

The Challenge of Using LLMs to Simulate Human Behavior: A Causal Inference Perspective

Author

Listed:
  • George Gui
  • Olivier Toubia

Abstract

Large Language Models (LLMs) have demonstrated impressive potential to simulate human behavior. Using a causal inference framework, we empirically and theoretically analyze the challenges of conducting LLM-simulated experiments, and explore potential solutions. In the context of demand estimation, we show that variations in the treatment included in the prompt (e.g., price of focal product) can cause variations in unspecified confounding factors (e.g., price of competitors, historical prices, outside temperature), introducing endogeneity and yielding implausibly flat demand curves. We propose a theoretical framework suggesting this endogeneity issue generalizes to other contexts and won't be fully resolved by merely improving the training data. Unlike real experiments where researchers assign pre-existing units across conditions, LLMs simulate units based on the entire prompt, which includes the description of the treatment. Therefore, due to associations in the training data, the characteristics of individuals and environments simulated by the LLM can be affected by the treatment assignment. We explore two potential solutions. The first specifies all contextual variables that affect both treatment and outcome, which we demonstrate to be challenging for a general-purpose LLM. The second explicitly specifies the source of treatment variation in the prompt given to the LLM (e.g., by informing the LLM that the store is running an experiment). While this approach only allows the estimation of a conditional average treatment effect that depends on the specific experimental design, it provides valuable directional results for exploratory analysis.

Suggested Citation

  • George Gui & Olivier Toubia, 2023. "The Challenge of Using LLMs to Simulate Human Behavior: A Causal Inference Perspective," Papers 2312.15524, arXiv.org.
  • Handle: RePEc:arx:papers:2312.15524
    as

    Download full text from publisher

    File URL: http://arxiv.org/pdf/2312.15524
    File Function: Latest version
    Download Restriction: no
    ---><---

    References listed on IDEAS

    as
    1. John J. Horton, 2023. "Large Language Models as Simulated Economic Agents: What Can We Learn from Homo Silicus?," NBER Working Papers 31122, National Bureau of Economic Research, Inc.
    2. Günter J. Hitsch & Ali Hortaçsu & Xiliang Lin, 2021. "Prices and promotions in U.S. retail markets," Quantitative Marketing and Economics (QME), Springer, vol. 19(3), pages 289-368, December.
    3. John J. Horton, 2023. "Large Language Models as Simulated Economic Agents: What Can We Learn from Homo Silicus?," Papers 2301.07543, arXiv.org.
    Full references (including those not matched with items on IDEAS)

    Most related items

    These are the items that most often cite the same works as this one and are cited by the same works as this one.
    1. Kevin Leyton-Brown & Paul Milgrom & Neil Newman & Ilya Segal, 2023. "Artificial Intelligence and Market Design: Lessons Learned from Radio Spectrum Reallocation," NBER Chapters, in: New Directions in Market Design, National Bureau of Economic Research, Inc.
    2. Zengqing Wu & Run Peng & Xu Han & Shuyuan Zheng & Yixin Zhang & Chuan Xiao, 2023. "Smart Agent-Based Modeling: On the Use of Large Language Models in Computer Simulations," Papers 2311.06330, arXiv.org, revised Dec 2023.
    3. Joshua C. Yang & Marcin Korecki & Damian Dailisan & Carina I. Hausladen & Dirk Helbing, 2024. "LLM Voting: Human Choices and AI Collective Decision Making," Papers 2402.01766, arXiv.org.
    4. Nir Chemaya & Daniel Martin, 2023. "Perceptions and Detection of AI Use in Manuscript Preparation for Academic Journals," Papers 2311.14720, arXiv.org, revised Jan 2024.
    5. Lijia Ma & Xingchen Xu & Yong Tan, 2024. "Crafting Knowledge: Exploring the Creative Mechanisms of Chat-Based Search Engines," Papers 2402.19421, arXiv.org.
    6. Ali Goli & Amandeep Singh, 2023. "Exploring the Influence of Language on Time-Reward Perceptions in Large Language Models: A Study Using GPT-3.5," Papers 2305.02531, arXiv.org, revised Jun 2023.
    7. Christoph Engel & Max R. P. Grossmann & Axel Ockenfels, 2023. "Integrating machine behavior into human subject experiments: A user-friendly toolkit and illustrations," Discussion Paper Series of the Max Planck Institute for Research on Collective Goods 2024_01, Max Planck Institute for Research on Collective Goods.
    8. Yiting Chen & Tracy Xiao Liu & You Shan & Songfa Zhong, 2023. "The emergence of economic rationality of GPT," Proceedings of the National Academy of Sciences, Proceedings of the National Academy of Sciences, vol. 120(51), pages 2316205120-, December.
    9. Jiafu An & Difang Huang & Chen Lin & Mingzhu Tai, 2024. "Measuring Gender and Racial Biases in Large Language Models," Papers 2403.15281, arXiv.org.
    10. Fulin Guo, 2023. "GPT in Game Theory Experiments," Papers 2305.05516, arXiv.org, revised Dec 2023.
    11. Felix Chopra & Ingar Haaland, 2023. "Conducting qualitative interviews with AI," CEBI working paper series 23-06, University of Copenhagen. Department of Economics. The Center for Economic Behavior and Inequality (CEBI).
    12. Siting Lu, 2024. "Strategic Interactions between Large Language Models-based Agents in Beauty Contests," Papers 2404.08492, arXiv.org.
    13. Keegan Harris & Nicole Immorlica & Brendan Lucier & Aleksandrs Slivkins, 2023. "Algorithmic Persuasion Through Simulation," Papers 2311.18138, arXiv.org, revised Apr 2024.
    14. Leland Bybee, 2023. "Surveying Generative AI's Economic Expectations," Papers 2305.02823, arXiv.org, revised May 2023.
    15. Emaad Manzoor & Nikhil Malik, 2023. "Designing Effective Music Excerpts," Papers 2309.14475, arXiv.org.
    16. Gary Charness & Brian Jabarian & John List, 2023. "Generation Next: Experimentation with AI," Artefactual Field Experiments 00777, The Field Experiments Website.
    17. Bauer, Kevin & Liebich, Lena & Hinz, Oliver & Kosfeld, Michael, 2023. "Decoding GPT's hidden "rationality" of cooperation," SAFE Working Paper Series 401, Leibniz Institute for Financial Research SAFE.
    18. Miguel Faria-e-Castro & Fernando Leibovici, 2023. "Artificial Intelligence and Inflation Forecasts," Working Papers 2023-015, Federal Reserve Bank of St. Louis, revised 26 Feb 2024.
    19. Zengqing Wu & Shuyuan Zheng & Qianying Liu & Xu Han & Brian Inhyuk Kwon & Makoto Onizuka & Shaojie Tang & Run Peng & Chuan Xiao, 2024. "Shall We Talk: Exploring Spontaneous Collaborations of Competing LLM Agents," Papers 2402.12327, arXiv.org.
    20. Yan Leng & Yuan Yuan, 2023. "Do LLM Agents Exhibit Social Behavior?," Papers 2312.15198, arXiv.org, revised Feb 2024.

    More about this item

    NEP fields

    This paper has been announced in the following NEP Reports:

    Statistics

    Access and download statistics

    Corrections

    All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:arx:papers:2312.15524. See general information about how to correct material in RePEc.

    If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about.

    If CitEc recognized a bibliographic reference but did not link an item in RePEc to it, you can help with this form .

    If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation.

    For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: arXiv administrators (email available below). General contact details of provider: http://arxiv.org/ .

    Please note that corrections may take a couple of weeks to filter through the various RePEc services.

    IDEAS is a RePEc service. RePEc uses bibliographic data supplied by the respective publishers.