Does Theory of Mind Improvement Really Benefit Human-AI Interactions? Empirical Findings from Interactive Evaluations
For researchers developing socially aware LLMs, this work highlights the critical gap between static ToM benchmarks and real-world interaction effectiveness, advocating for interaction-based evaluations.
The paper investigates whether improvements in Theory of Mind (ToM) for LLMs, as measured by static benchmarks, translate to better human-AI interactions. Through interactive evaluations with real-world datasets and user studies, they found that static benchmark gains do not consistently lead to improved dynamic interaction performance.
Improving the Theory of Mind (ToM) capability of Large Language Models (LLMs) is crucial for effective social interactions between these AI models and humans. However, the existing benchmarks often measure ToM capability improvement through story-reading, multiple-choice questions from a third-person perspective, while ignoring the first-person, dynamic, and open-ended nature of human-AI (HAI) interactions. To directly examine how ToM improvement techniques benefit HAI interactions, we first proposed the new paradigm of interactive ToM evaluation with both perspective and metric shifts. Next, following the paradigm, we conducted a systematic study of four representative ToM enhancement techniques using both four real-world datasets and a user study, covering both goal-oriented tasks (e.g., coding, math) and experience-oriented tasks (e.g., counseling). Our findings reveal that improvements on static benchmarks do not always translate to better performance in dynamic HAI interactions. This paper offers critical insights into ToM evaluation, showing the necessity of interaction-based assessments in developing next-generation, socially aware LLMs for HAI symbiosis.