Study finds curiosity-oriented LLM questions remain uncommon
Original title: What Makes a Question Curious? Measuring Model Curiosity in Human-LLM Interactions
Across four datasets, such questions were 17-28% of all questions and tied to richer later user replies.
Researchers define model curiosity as epistemic question-asking that surfaces gaps, assumptions, and alternatives, distinct from generic questioning. Across four conversation datasets, curiosity-oriented questions were uncommon but associated with more exploratory user responses, with patterns varying by dataset. A human study found curiosity can encourage exploration, especially in advice, creative, and research contexts.
Unconfirmed: Usefulness conclusions rest on one human study, and dataset patterns were not uniform.
- Curiosity-oriented questions accounted for 17-28% of all questions.
- Associations with later responses varied across the four datasets.
- Human evaluators saw the strongest benefit in advice, creative, and research settings.
- The authors argue selective questioning could support human-AI collaboration.
From the original article
1 min read at arxiv.orgLarge language models increasingly act as partners in everyday communication. In these interactions, useful assistance may require not only answering, but also asking the right questions that surface knowledge gaps, assumptions, and alternatives.
By Angana Borah · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.