arXiv AI

When Agents Say One Thing and Do Another: Validating Elicited Beliefs from LLMs

The paper introduces a decision‑theoretic framework that elicits both probability judgments and decisions from large language models (LLMs) to test whether their reported beliefs are consistent with their actions. It shows that this framework yields empirically testable conditions without assuming a specific utility function. In clinical diagnosis simulations, the authors find that while LLMs’ reported beliefs are not perfect reflections of the information in their decisions, the discrepancies are small for the strongest models.

arXiv Machine Learning
Sep 10

Can Revealed Preferences Clarify LLM Alignment and Steering?

The paper proposes an empirical pipeline to estimate the preferences that a large language model (LLM) implicitly optimizes by combining the model’s probability distribution over unknowns with its chosen action, and fitting a discrete choice model to recover the underlying cost function. This revealed-preference framework enables rigorous assessment of whether LLMs act consistently toward a goal, can articulate objectives that align with their decision policy, and can be steered by prompting to follow a user-specified cost function. Experiments across four medical diagnosis domains and various frontier and open-source models show that while many LLMs exhibit internal coherence, they still struggle to accurately report or adopt preferences when guided by users.

By Khurram Yamin, Jingjing Tang, Eric Horvitz, Bryan Wilder
arXiv AI
Sep 10

Beliefs and Behavior in Language Models

arXiv:2609.07943v1 Announce Type: new Abstract: There is significant uncertainty about whether abstractions like beliefs or desires usefully describe the behavior of large language models (LLMs). In...

By Alex Smolin, Bryan Wilder
arXiv Computation and Language
Sep 1

CARE: Privacy-Compliant Agentic Reasoning with Evidence Discordance

The paper introduces MIMIC-DOS, a dataset derived from MIMIC-IV that focuses on ICU cases where patient symptoms and medical signs are discordant. It presents CARE, a privacy‑compliant multi‑stage agentic reasoning framework that uses a proprietary LLM to generate structured categories and transitions, while a local LLM performs evidence acquisition and decision‑making. In retrospective evaluations on MIMIC‑DOS, CARE outperforms other LLMs and agentic workflows, demonstrating stronger handling of conflicting clinical evidence while preserving patient privacy.

By Haochen Liu, Weien Li, Rui Song, Zeyu Li, Chun Jason Xue, Xiao-Yang Liu, Sam Nallaperuma-Herzberg, Xue Liu, Ye Yuan
arXiv AI
Jul 28

Rethinking Prospect Theory for LLMs: Revealing the Instability of Decision-Making under Epistemic Uncertainty

arXiv:2508. 08992v4 Announce Type: replace Abstract: Real-world decision-making often involves uncertainty expressed in linguistic rather than numerical terms, and Prospect Theory (PT) provides a classic framework for modeling human behavior under such uncertainty.

By Rui Wang, Qihan Lin, Jiayu Liu, Qing Zong, Tianshi Zheng, Dadi Guo, Haochen Shi, Peixuan Han, Weiqi Wang, Yangqiu Song
arXiv AI
6d ago

Can LLMs Follow Medical Expert Logic? A Benchmark for Hierarchical Logical Consistency in Risk-of-Bias Assessment

LogiMed‑RoB is a new benchmark that tests large language models (LLMs) on hierarchical logical consistency in medical risk‑of‑bias assessments, using 860 randomized controlled trials and 14,820 queries based on Cochrane Risk of Bias 2.0 expert logic. The benchmark evaluates models across four dimensions—Atomic Consistency, Domain Consistency, Aggregation Consistency, and Evidential Faithfulness—revealing a catastrophic error‑compounding effect where high atomic accuracy does not translate to end‑to‑end consistency. Experiments on ten state‑of‑the‑art LLMs show that even top models can fail to deduce correct outcomes in a significant portion of cases, highlighting a gap between evidence retrieval and reasoning. whyItMatters":"The study shows that high outcome accuracy can mask critical reasoning flaws, emphasizing the need for white‑box logical verification before deploying LLMs in clinical settings."

By Jiayu Huang, Zichen Tang, Qianhui Ling, Zemin Kuang, Haihong E
Hugging Face Trending Papers
Sep 10

Can LLMs Follow Medical Expert Logic? A Benchmark for Hierarchical Logical Consistency in Risk-of-Bias Assessment

The paper introduces LogiMed‑RoB, a benchmark that tests large language models (LLMs) on hierarchical logical consistency in medical risk‑of‑bias assessments, using 860 randomized controlled trials and 14,820 queries based on Cochrane RoB 2.0 expert logic. It evaluates models across four dimensions—Atomic Consistency, Domain Consistency, Aggregation Consistency, and Evidential Faithfulness—revealing a severe Error Compounding Effect where high atomic accuracy does not translate to end‑to‑end consistency. Experiments on ten state‑of‑the‑art LLMs show that even top performers can collapse to 45.13% overall consistency, with some models nearly failing entirely, and that many models struggle to deduce correct outcomes from retrieved evidence.