Saved in:
Bibliographic Details
Main Authors: Hutson, Dylan, Vennemeyer, Daniel, Deshmukh, Aneesh, Zhan, Justin, Jiang, Tianyu
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.19593
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915510702047232
author Hutson, Dylan
Vennemeyer, Daniel
Deshmukh, Aneesh
Zhan, Justin
Jiang, Tianyu
author_facet Hutson, Dylan
Vennemeyer, Daniel
Deshmukh, Aneesh
Zhan, Justin
Jiang, Tianyu
contents We introduce GuessingGame, a protocol for evaluating large language models (LLMs) as strategic question-askers in open-ended, open-domain settings. A Guesser LLM identifies a hidden object by posing free-form questions to an Oracle without predefined choices or candidate lists. To measure question quality, we propose two information gain (IG) metrics: a Bayesian method that tracks belief updates over semantic concepts using LLM-scored relevance, and an entropy-based method that filters candidates via ConceptNet. Both metrics are model-agnostic and support post hoc analysis. Across 858 games with multiple models and prompting strategies, higher IG strongly predicts efficiency: a one-standard-deviation IG increase reduces expected game length by 43\%. Prompting constraints guided by IG, such as enforcing question diversity, enable weaker models to significantly improve performance. These results show that question-asking in LLMs is both measurable and improvable, and crucial for interactive reasoning.
format Preprint
id arxiv_https___arxiv_org_abs_2509_19593
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle GuessingGame: Measuring the Informativeness of Open-Ended Questions in Large Language Models
Hutson, Dylan
Vennemeyer, Daniel
Deshmukh, Aneesh
Zhan, Justin
Jiang, Tianyu
Computation and Language
Artificial Intelligence
We introduce GuessingGame, a protocol for evaluating large language models (LLMs) as strategic question-askers in open-ended, open-domain settings. A Guesser LLM identifies a hidden object by posing free-form questions to an Oracle without predefined choices or candidate lists. To measure question quality, we propose two information gain (IG) metrics: a Bayesian method that tracks belief updates over semantic concepts using LLM-scored relevance, and an entropy-based method that filters candidates via ConceptNet. Both metrics are model-agnostic and support post hoc analysis. Across 858 games with multiple models and prompting strategies, higher IG strongly predicts efficiency: a one-standard-deviation IG increase reduces expected game length by 43\%. Prompting constraints guided by IG, such as enforcing question diversity, enable weaker models to significantly improve performance. These results show that question-asking in LLMs is both measurable and improvable, and crucial for interactive reasoning.
title GuessingGame: Measuring the Informativeness of Open-Ended Questions in Large Language Models
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2509.19593