Can Large Language Models generalize analogy solving like children can?
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914076891807744 |
|---|---|
| author | Stevenson, Claire E. Pafford, Alexandra van der Maas, Han L. J. Mitchell, Melanie |
| author_facet | Stevenson, Claire E. Pafford, Alexandra van der Maas, Han L. J. Mitchell, Melanie |
| contents | In people, the ability to solve analogies such as "body : feet :: table : ?" emerges in childhood, and appears to transfer easily to other domains, such as the visual domain "( : ) :: < : ?". Recent research shows that large language models (LLMs) can solve various forms of analogies. However, can LLMs generalize analogy solving to new domains like people can? To investigate this, we had children, adults, and LLMs solve a series of letter-string analogies (e.g., a b : a c :: j k : ?) in the Latin alphabet, in a near transfer domain (Greek alphabet), and a far transfer domain (list of symbols). Children and adults easily generalized their knowledge to unfamiliar domains, whereas LLMs did not. This key difference between human and AI performance is evidence that these LLMs still struggle with robust human-like analogical transfer. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_02348 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Can Large Language Models generalize analogy solving like children can? Stevenson, Claire E. Pafford, Alexandra van der Maas, Han L. J. Mitchell, Melanie Artificial Intelligence Computation and Language Human-Computer Interaction In people, the ability to solve analogies such as "body : feet :: table : ?" emerges in childhood, and appears to transfer easily to other domains, such as the visual domain "( : ) :: < : ?". Recent research shows that large language models (LLMs) can solve various forms of analogies. However, can LLMs generalize analogy solving to new domains like people can? To investigate this, we had children, adults, and LLMs solve a series of letter-string analogies (e.g., a b : a c :: j k : ?) in the Latin alphabet, in a near transfer domain (Greek alphabet), and a far transfer domain (list of symbols). Children and adults easily generalized their knowledge to unfamiliar domains, whereas LLMs did not. This key difference between human and AI performance is evidence that these LLMs still struggle with robust human-like analogical transfer. |
| title | Can Large Language Models generalize analogy solving like children can? |
| topic | Artificial Intelligence Computation and Language Human-Computer Interaction |
| url | https://arxiv.org/abs/2411.02348 |