Theory of Mind (ToM) originates from psychology and refers to the ability to infer others’ internal mental states, such as emotions, beliefs, knowledge, desires, and intentions. It is a fundamental basis of human social interaction. As AI becomes increasingly widespread, human-AI collaboration is expected to become the norm. Equipping AI with the social intelligence enabled by ToM will therefore be a key technology for facilitating effective collaboration. In this talk, we will discuss machine ToM across three stages of AI system development: training, evaluation, and applications. Specifically, we will highlight cost-efficient training, reliable evaluation, and broad application scenarios, including mental manipulation detection, gaming, and autonomous driving. Overall, this talk discusses the development of machine ToM and its potential contribution to socially intelligent and trustworthy AI systems.

Shao-Yuan Lo is an Assistant Professor (Yushan Young Fellow) at National Taiwan University (NTU). Prior to joining NTU, he was a Research Scientist at Honda Research Institute USA. He received his Ph.D. from Johns Hopkins University in 2023 and his M.S. and B.S. degrees from National Chiao Tung University in 2019 and 2017, respectively. His recent research focuses on Multimodal Foundation Models and Trustworthy AI. He has first- or corresponding-authored over 20 publications in venues such as IEEE T-PAMI, IEEE T-IP, IJCV, ICML (Spotlight), CVPR (Highlight), and ECCV. He won the Outstanding Reviewer at CVPR 2025 and the Best Paper Award at ACM Multimedia Asia 2019.