Theory of Mind
Theory of Mind (ToM) research investigates the ability to understand and predict others' mental states, a crucial aspect of social intelligence. Current research focuses on enhancing ToM capabilities in large language models (LLMs) through methods like prompting techniques (e.g., chain-of-thought, metacognitive prompting) and incorporating external symbolic reasoning systems. This work utilizes various benchmarks, including those based on narratives, visual data, and real-world interactions, to evaluate and improve LLM performance on ToM tasks. The advancements in this field have significant implications for developing more robust and socially intelligent AI systems, with applications ranging from human-robot collaboration to improved educational technologies.
Papers
Views Are My Own, but Also Yours: Benchmarking Theory of Mind Using Common Ground
Adil Soubki, John Murzaku, Arash Yousefi Jordehi, Peter Zeng, Magdalena Markowska, Seyed Abolghasem Mirroshandel, Owen Rambow
PHAnToM: Persona-based Prompting Has An Effect on Theory-of-Mind Reasoning in Large Language Models
Fiona Anting Tan, Gerard Christopher Yeo, Kokil Jaidka, Fanyou Wu, Weijie Xu, Vinija Jain, Aman Chadha, Yang Liu, See-Kiong Ng