Agents Explore but Agents Ignore: LLMs Lack Environmental Curiosity

TL;DR

This paper demonstrates that current LLM-based agents lack environmental curiosity, failing to exploit unexpected but relevant information, which limits their ability to adapt and maximize utility in various tasks.

Contribution

It reveals the gap between discovering and exploiting solutions in LLM agents and identifies factors affecting environmental curiosity, proposing areas for improvement.

Findings

Agents discover solutions in 79-81% of Terminal-Bench runs

Agents exploit solutions in only 37-50% of Terminal-Bench runs

Agents rarely exploit solutions in AppWorld despite clear documentation

Abstract

LLM-based agents are assumed to integrate environmental observations into their reasoning: discovering highly relevant but unexpected information should naturally lead to a model exploiting its own discoveries. We show that this assumption is false for current LLM-based agents, which struggle to reflect or react to unexpected information. Across three benchmarks (Terminal-Bench, SWE-Bench, AppWorld), we inject complete task solutions into the agent environments to deliberately expose a task's solution to a model. While agents discover these solutions on Terminal-Bench in 79-81% of runs, they interact, or exploit, them in only 37-50% of cases. This gap is starkest in AppWorld: agents see documentation stating that a command "returns the complete solution to this task" in over 90% of attempts but exploit this in fewer than 7% of trials. We show that agents lack what we call environmental curiosity: the capability to recognize and investigate unexpected but relevant observations in response to environmental stimuli. We identify three main factors influencing environmental curiosity: available tools in the agent scaffold, test-time compute, and training data distribution. Our findings identify configurations that maximize curiosity also achieve the best performance on the unmodified benchmarks. Yet even jointly optimized agents still ignore discovered solutions in the majority of trials: current agents use the environment to fetch expected information, but not to revise their strategy or maximally exploit useful stimuli.