Altruistic behavior in generative artificial intelligence: a comparative study using Dictator Game experiments
Article Sidebar
Main Article Content
Abstract
This paper examines the growing integration of artificial intelligence (AI) in various social domains, with a focus on altruistic behavior exhibited by large language models (LLMs) such as GPT-3.5-turbo. As these models evolve and gain sophistication, there is a need to investigate whether they can replicate the complexity of human behavior in moral decision-making, particularly in situations involving altruism. The study uses the Dictator Game, a classical experiment in behavioral economics, to analyze how LLMs respond in scenarios where an agent decides how to divide a sum of money between themselves and an anonymous recipient without expecting any reward. The key research question is whether different versions of GPT-3.5-turbo exhibit variation in altruistic decision-making. The comparison of GPT-3.5-turbo’s decisions with human behavior offers valuable insights into AI-human interaction in moral contexts.
Article Details
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Los autores conservan los derechos de autor y ceden a la revista el derecho de la primera publicación y pueda editarlo, reproducirlo, distribuirlo, exhibirlo y comunicarlo en el país y en el extranjero mediante medios impresos y electrónicos. Asimismo, asumen el compromiso sobre cualquier litigio o reclamación relacionada con derechos de propiedad intelectual, exonerando de responsabilidad a la Editorial Tecnológica de Costa Rica. Además, se establece que los autores pueden realizar otros acuerdos contractuales independientes y adicionales para la distribución no exclusiva de la versión del artículo publicado en esta revista (p. ej., incluirlo en un repositorio institucional o publicarlo en un libro) siempre que indiquen claramente que el trabajo se publicó por primera vez en esta revista.
References
[1] Ma, J., “Can Machines Think Like Humans? A Behavioral Evaluation of LLM-Agents in Dictator Games,” Journal of Behavioral and Experimental Economics, 2024. doi : 10.1016/j.joep.2024.103107
[2] Mei, Q., et al., “A Turing Test of Whether AI Chatbots Are Behaviorally Similar to Humans,” Proceedings of the National Academy of Sciences, vol. 121, no. 9, 2024. doi: 10.1073/pnas.2313925121.
[3] Brookins, P., & DeBacker, J., “Playing Games With GPT,” SSRN Electronic Journal, 2024. doi: 10.2139/ssrn.4493398.
[4] Engel, C., “Dictator Games: A Meta Study,” Experimental Economics, vol. 14, no. 4, pp. 583–610, 2011. doi: 10.1007/s10683-011-9283-7.
[5] E.-M. Schmidt, S. Bonati, N. Köbis, and I. Soraperra, “GPT-3.5 altruistic advice is sensitive to reciprocal concerns but not to strategic risk,” Sci Rep, vol. 14, no. 1, Sep. 2024, doi: 10.1038/s41598-024-73306-x
[6] I. Thielmann, R. Böhm, M. Ott, and B. E. Hilbig, “Economic Games: An Introduction and Guide for Research,” Collabra: Psychology, vol. 7, no. 1, art. 19004, 2021. doi: 10.1525/collabra.19004.
[7] A. C. C. de Oliveira, “Altruism in Economic Decision-Making,” Journal of Behavioral Economics, vol. 12, no. 3, pp. 45-60, 2020.
[8] T. Johnson y A. Obradovich, “Altruism and Selfishness in Believable Game Agents: Deep Reinforcement Learning in Modified Dictator Games,” Artificial Intelligence Review, vol. 54, no. 2, pp. 123-145, 2021.
[9] A. Radford, J. Wu, R. Child, D. Luan, D. Amodei, I, Sustskever, “Language Models are Unsupervised Multitask Learners,” OpenAI, 2019.
[10] OpenAI, “GPT-3.5 Turbo model documentation,” 2023. [Online]. Available: https://platform.openai.com/docs [Accessed:9 de Agosto de 2024].
[11] OpenAI, “Create chat”. [Online]. Available: API Reference - OpenAI API [Accessed: 14 de Agosto de 2024].
[12] T. Johnson and N. Obradovich, “Testing for completions that simulate altruism in early language models,” Nat Hum Behav, vol. 9, no. 9, pp. 1861–1870, Jul. 2025, doi: 10.1038/s41562-025-02258-7.