Welcome to the Computation & Cognition Lab at KAIST AI. Our mission is to advance AI by combining computational methods with insights derived from human cognition and development, particularly those rooted in social cognition. We pursue this goal through two complementary directions: machine social cognition and social world modeling. In machine social cognition, we draw inspiration from humans’ remarkable ability to learn efficiently from others and share knowledge through interaction. We explore how machines can improve their reasoning and learning through computation, data, memory, and feedback from others or themselves. In social world modeling, we study how machines can understand people’s mental states, social contexts, norms, and embodied interactions, with applications to human behavior simulation and physical AI.

Our current research interests include (but are not limited to):

• Reasoning and inference across cognitive and social domains, including multimodal settings.
• Scalable and efficient learning through distillation, data synthesis, selection, and curriculum.
• Systematic evaluation of cognitive behaviors in foundation models.
• Large-scale data synthesis for diverse purposes and domains.
• Agents engaging in interactions in various environments.
• Simulation of individual and collective human behavior.
• Social world models for virtual agents and physical AI.