Testing theory of mind in large language models and humans.

Autor: Strachan JWA; Department of Neurology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany. james.wa.strachan@gmail.com., Albergo D; Cognition, Motion and Neuroscience, Italian Institute of Technology, Genoa, Italy.; Center for Mind/Brain Sciences, University of Trento, Rovereto, Italy., Borghini G; Cognition, Motion and Neuroscience, Italian Institute of Technology, Genoa, Italy., Pansardi O; Department of Neurology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany.; Cognition, Motion and Neuroscience, Italian Institute of Technology, Genoa, Italy.; Department of Psychology, University of Turin, Turin, Italy., Scaliti E; Department of Neurology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany.; Cognition, Motion and Neuroscience, Italian Institute of Technology, Genoa, Italy.; Department of Management, 'Valter Cantino', University of Turin, Turin, Italy.; Human Science and Technologies, University of Turin, Turin, Italy., Gupta S; Alien Technology Transfer Ltd, London, UK., Saxena K; Alien Technology Transfer Ltd, London, UK., Rufo A; Alien Technology Transfer Ltd, London, UK., Panzeri S; Institute for Neural Information Processing, Center for Molecular Neurobiology, University Medical Center Hamburg- Eppendorf, Hamburg, Germany., Manzi G; Alien Technology Transfer Ltd, London, UK., Graziano MSA; Princeton Neuroscience Institute, Princeton University, Princeton, NJ, USA., Becchio C; Department of Neurology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany. c.becchio@uke.de.; Cognition, Motion and Neuroscience, Italian Institute of Technology, Genoa, Italy. c.becchio@uke.de.
Jazyk: angličtina
Zdroj: Nature human behaviour [Nat Hum Behav] 2024 Jul; Vol. 8 (7), pp. 1285-1295. Date of Electronic Publication: 2024 May 20.
DOI: 10.1038/s41562-024-01882-z
Abstrakt: At the core of what defines us as humans is the concept of theory of mind: the ability to track other people's mental states. The recent development of large language models (LLMs) such as ChatGPT has led to intense debate about the possibility that these models exhibit behaviour that is indistinguishable from human behaviour in theory of mind tasks. Here we compare human and LLM performance on a comprehensive battery of measurements that aim to measure different theory of mind abilities, from understanding false beliefs to interpreting indirect requests and recognizing irony and faux pas. We tested two families of LLMs (GPT and LLaMA2) repeatedly against these measures and compared their performance with those from a sample of 1,907 human participants. Across the battery of theory of mind tests, we found that GPT-4 models performed at, or even sometimes above, human levels at identifying indirect requests, false beliefs and misdirection, but struggled with detecting faux pas. Faux pas, however, was the only test where LLaMA2 outperformed humans. Follow-up manipulations of the belief likelihood revealed that the superiority of LLaMA2 was illusory, possibly reflecting a bias towards attributing ignorance. By contrast, the poor performance of GPT originated from a hyperconservative approach towards committing to conclusions rather than from a genuine failure of inference. These findings not only demonstrate that LLMs exhibit behaviour that is consistent with the outputs of mentalistic inference in humans but also highlight the importance of systematic testing to ensure a non-superficial comparison between human and artificial intelligences.
(© 2024. The Author(s).)
Databáze: MEDLINE
Externí odkaz: