The Curious Language Model: Strategic Test-Time Information Acquisition

TL;DR

This paper introduces CuriosiTree, a heuristic test-time policy for large language models that strategically acquires information cost-effectively, improving decision accuracy in clinical diagnosis simulations.

Contribution

The paper presents CuriosiTree, a novel greedy tree search-based method for zero-shot, cost-aware information acquisition in large language models.

Findings

Outperforms baseline strategies in clinical diagnosis accuracy

Enables cost-effective integration of diverse information sources

Improves decision confidence with strategic action selection

Abstract

Decision-makers often possess insufficient information to render a confident decision. In these cases, the decision-maker can often undertake actions to acquire the necessary information about the problem at hand, e.g., by consulting knowledgeable authorities or by conducting experiments. Importantly, different levers of information acquisition come with different costs, posing the challenge of selecting the actions that are both informative and cost-effective. In this work, we propose CuriosiTree, a heuristic-based, test-time policy for zero-shot information acquisition in large language models (LLMs). CuriosiTree employs a greedy tree search to estimate the expected information gain of each action and strategically chooses actions based on a balance of anticipated information gain and associated cost. Empirical validation in a clinical diagnosis simulation shows that CuriosiTree enables cost-effective integration of heterogenous sources of information, and outperforms baseline action selection strategies in selecting action sequences that enable accurate diagnosis.