Assistive AI with Artificial Theories of Mind and Body

Project Description

Although providing appropriate assistance is commonplace task for humans, it remains a significant challenge for artificial intelligence. We would like our artificial agents to help us in effective and human-interpretable ways, but current AI systems lack a good understanding of human users. One reason for this is that good human models are difficult to develop. Current theory-driven approaches in the cognitive sciences focus on causal explanations for human behaviour and face difficulties scaling up beyond simple scenarios. On the other hand, data-driven deep learning can learn high-performance complex models, but typically requires large amounts of data to generalize well. We intend to build novel hybrid techniques that learn flexible deep models, yet are able to leverage prior knowledge that is easy to specify and interpret. We envision our methods would yield accurate human models with less data than existing approaches, and plan to demonstrate that the use of human models leads to better assistive AI.

Research Technical Area

  • Cognitive modelling and systems
  • Machine learning
  • Robotics

Benefits to the society

Achieving our research objectives will have broad scientific significance for artificial intelligence, machine learning, and human-AI interaction. A Theory of Mind and Body is essential for human social interactions— we use theories of Mind/Body to explain and predict behaviour— and artificial versions will enable AI to engage in natural cognitive and physical interactions with people. As an example, our methods could enhance intelligent assistance for health applications, e.g., advising diabetics on food/exercise. More broadly, artificial Theories of Mind/Body may form a core component for enabling machines to infer social and moral/ethical norms; understanding the effect of one’s actions on sentient entities is key for building ethical machines.

Project’s Publications

Team’s Principal Investigator

Assistant Professor Harold Soh Soon Hong

National University of Singapore

Harold Soh is Assistant Professor at the Department of Computer Science at the National University of Singapore, where he directs the Collaborative Learning and Adaptive Robots (CLeAR) group. His research interests are in human-AI/robot interaction, machine learning, and robotics. Harold obtained his PhD from Imperial College London where he developed online learning for assistive robotics. He currently focusses on developing trustworthy robots that interact fluently with people. His research work has been nominated for awards at top robotics conferences (RSS, HRI, IROS).


Recent Notable Awards

  • Best Paper Award Finalist, Robotics Science and Systems (RSS), 2018
  • Best Paper Award Finalist, ACM/IEEE Human Robot Interaction (HRI), 2018
  • Best Long Paper Award Runner-up, ACM Recommender Systems (RecSys), 2018

The Team


Dr. Desmond ONG, National University of Singapore
Research Focus: Computational cognitive models of reasoning, Affective computation, Probabilistic programming