Causality Elicitation from Large Language Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908866159050752 |
|---|---|
| author | Kameyama, Takashi Kato, Masahiro Hio, Yasuko Takano, Yasushi Minakawa, Naoto |
| author_facet | Kameyama, Takashi Kato, Masahiro Hio, Yasuko Takano, Yasushi Minakawa, Naoto |
| contents | Large language models (LLMs) are trained on enormous amounts of data and encode knowledge in their parameters. We propose a pipeline to elicit causal relationships from LLMs. Specifically, (i) we sample many documents from LLMs on a given topic, (ii) we extract an event list from from each document, (iii) we group events that appear across documents into canonical events, (iv) we construct a binary indicator vector for each document over canonical events, and (v) we estimate candidate causal graphs using causal discovery methods. Our approach does not guarantee real-world causality. Rather, it provides a framework for presenting the set of causal hypotheses that LLMs can plausibly assume, as an inspectable set of variables and candidate graphs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_04276 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Causality Elicitation from Large Language Models Kameyama, Takashi Kato, Masahiro Hio, Yasuko Takano, Yasushi Minakawa, Naoto Machine Learning Artificial Intelligence Computation and Language Econometrics Large language models (LLMs) are trained on enormous amounts of data and encode knowledge in their parameters. We propose a pipeline to elicit causal relationships from LLMs. Specifically, (i) we sample many documents from LLMs on a given topic, (ii) we extract an event list from from each document, (iii) we group events that appear across documents into canonical events, (iv) we construct a binary indicator vector for each document over canonical events, and (v) we estimate candidate causal graphs using causal discovery methods. Our approach does not guarantee real-world causality. Rather, it provides a framework for presenting the set of causal hypotheses that LLMs can plausibly assume, as an inspectable set of variables and candidate graphs. |
| title | Causality Elicitation from Large Language Models |
| topic | Machine Learning Artificial Intelligence Computation and Language Econometrics |
| url | https://arxiv.org/abs/2603.04276 |