EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Large language models show amplified cognitive biases in moral decision-making

Vanessa Cheung iD, Maximilian Maier iD, Falk Lieder iD

Proceedings of the National Academy of Sciences · 24.06.2025 · 122 (25)

DOI: 10.1073/pnas.2412015122 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 20.06.2025. Verfasst von Vanessa Cheung und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Experimental Psychology, University College London, London WC1H 0AP, United Kingdom. Hauptthemen laut MEDLINE: Decision Making, Morals, Cognition, Language. Die Arbeit wurde bisher 64-mal zitiert (Crossref) und nennt 117 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: Forethought Foundation for Global Priorities Research.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

As large language models (LLMs) become more widely used, people increasingly rely on them to make or advise on moral decisions. Some researchers even propose using LLMs as participants in psychology experiments. It is, therefore, important to understand how well LLMs make moral decisions and how they compare to humans. We investigated these questions by asking a range of LLMs to emulate or advise on people’s decisions in realistic moral dilemmas. In Study 1, we compared LLM responses to those of a representative U.S. sample ( N = 285) for 22 dilemmas, including both collective action problems that pitted self-interest against the greater good, and moral dilemmas that pitted utilitarian cost–benefit reasoning against deontological rules. In collective action problems, LLMs were more altruistic than participants. In moral dilemmas, LLMs exhibited stronger omission bias than participants: They usually endorsed inaction over action. In Study 2 ( N = 474, preregistered), we replicated this omission bias and documented an additional bias: Unlike humans, most LLMs were biased toward answering “no” in moral dilemmas, thus flipping their decision/advice depending on how the question is worded. In Study 3 ( N = 491, preregistered), we replicated these biases in LLMs using everyday moral dilemmas adapted from forum posts on Reddit. In Study 4, we investigated the sources of these biases by comparing models with and without fine-tuning, showing that they likely arise from fine-tuning models for chatbot applications. Our findings suggest that uncritical reliance on LLMs’ moral decisions and advice could amplify human biases and introduce potentially problematic biases.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Decision MakingMoralsCognitionLanguageHumansMaleFemaleAdultBiasYoung AdultLarge Language Models

Stichwörter der Autor:innen

ethical AIlarge language modelsmoral decision-makingmoral dilemmas

Autor:innen und Institutionen

  • Department of Experimental Psychology, University College London, London WC1H 0AP, United Kingdom.
  • Department of Psychology, University of California, Los Angeles, CA 90095.

ORCID-Profile: Vanessa Cheung, Maximilian Maier, Falk Lieder

Förderung

  • Forethought Foundation for Global Priorities Research

Zitieren

APA

Cheung, V., Maier, M., & Lieder, F. (2025). Large language models show amplified cognitive biases in moral decision-making. Proceedings of the National Academy of Sciences, 122(25). https://doi.org/10.1073/pnas.2412015122

Vancouver

Cheung V, Maier M, Lieder F. Large language models show amplified cognitive biases in moral decision-making. Proceedings of the National Academy of Sciences. 2025;122(25). doi:10.1073/pnas.2412015122

BibTeX
@article{Cheung2025,
  author = {Cheung, Vanessa and Maier, Maximilian and Lieder, Falk},
  title = {Large language models show amplified cognitive biases in moral decision-making},
  journal = {Proceedings of the National Academy of Sciences},
  year = {2025},
  volume = {122},
  number = {25},
  doi = {10.1073/pnas.2412015122}
}
Kostenlos starten →