FANToM: A Benchmark for Stress-testing Machine Theory of Mind in Interactions
This addresses the need for better evaluation of ToM in AI systems, particularly for interactive contexts, though it is incremental as it builds on existing ToM benchmarks by adding interactivity.
The authors tackled the problem of evaluating theory of mind (ToM) in large language models by introducing FANToM, a benchmark for stress-testing ToM in interactive, information-asymmetric conversations, and found that state-of-the-art LLMs perform significantly worse than humans, even with advanced techniques like chain-of-thought reasoning or fine-tuning.
Theory of mind (ToM) evaluations currently focus on testing models using passive narratives that inherently lack interactivity. We introduce FANToM, a new benchmark designed to stress-test ToM within information-asymmetric conversational contexts via question answering. Our benchmark draws upon important theoretical requisites from psychology and necessary empirical considerations when evaluating large language models (LLMs). In particular, we formulate multiple types of questions that demand the same underlying reasoning to identify illusory or false sense of ToM capabilities in LLMs. We show that FANToM is challenging for state-of-the-art LLMs, which perform significantly worse than humans even with chain-of-thought reasoning or fine-tuning.