Analogical reasoning is a hallmark of human intelligence, as it enables us to flexibly solve new problems without extensive practice. By using a wide range of tests, we demonstrate that GPT-3, a large-scale artificial intelligence language model, is capable of solving difficult analogy problems at a level comparable to human performance.
This is a preview of subscription content, access via your institution
Relevant articles
Open Access articles citing this article.
-
Who is the human in the machine? Releasing the human–machine metaphor from its cultural roots can increase innovation and equity in AI
AI and Ethics Open Access 19 December 2023
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$29.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 digital issues and online access to articles
$119.00 per year
only $9.92 per issue
Buy this article
- Purchase on Springer Link
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
References
Holyoak, K.J. in Oxford Handbook of Thinking and Reasoning (eds Holyoak, K. J. & Morrison, R. G.) 234–259 (Oxford Univ. Press, 2012). A book chapter that summarizes work in cognitive science on analogical reasoning.
Brown, T. et al. Language models are few-shot learners. In Adv. Neural Information Processing Systems 33 (eds Larochelle, H. et al.) 1877–1901 (Curran Associates, 2020). This paper describes GPT-3, the AI system that was evaluated in the present work.
Raven, J. C. Progressive Matrices: A Perceptual Test of Intelligence, Individual Form (Lewis Raven, 1938). A visual analogy problem set that is commonly used as a test of problem-solving skills.
Lake, B. M. et al. Building machines that learn and think like people. Behav. Brain Sci. 40, E253 (2017). A review and perspective that characterizes some limitations of deep learning systems.
Mitchell, M. Abstraction and analogy-making in artificial intelligence. Ann. NY Acad. Sci. 1505, 79–101 (2021). A review that summarizes work in AI on analogical reasoning.
Lu, H., Ichien, N. & Holyoak, K. J. Probabilistic analogical mapping with semantic relation networks. Psychol. Rev. 129, 1078 (2022). An example of work that combines deep learning with structured reasoning operations.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This is a summary of: Webb, T. et al. Emergent analogical reasoning in large language models. Nat. Hum. Behav. https://doi.org/10.1038/s41562-023-01659-w (2023).
Rights and permissions
About this article
Cite this article
Large-scale AI language systems display an emergent ability to reason by analogy. Nat Hum Behav 7, 1426–1427 (2023). https://doi.org/10.1038/s41562-023-01671-0
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41562-023-01671-0
