Title: Context Synchronization for Collaborative and Personalized AI

Speaker: Dr. Justin Cho @ University of Southern California

Time : 10:30 - 11:30, Nov 19th, 2025

Location: Online

https://hli.skku.edu/InvitedTalk251119 

Language: English speech & English slides

Abstract: 

Shared contextual understanding is fundamental to effective human communication. It conditions how meaning is interpreted, even when meaning is only partially specified. Although recent advances in conversational AI have transformed how we access information and interact with machines, the lack of this shared contextual understanding and the ability to navigate under such conditions are the key barriers for current systems to become effective human-centered collaborators. In this thesis, I present my research for overcoming these challenges through a framework of context synchronization, which identifies and bridges the contextual gaps in human-AI interaction. Specifically, I focus on (i) evaluating grounding gaps in language models on collaborative tasks, providing insights into their reliability for intervening on real-world interactions, (ii) simulating complex contexts for outcome-oriented optimization of open-ended tasks, and (iii) expanding contextual understanding beyond a textual interface to broader human-AI communication modalities. Finally, I discuss future directions for context synchronization to enable the next generation of collaborative and personalized AI agents, including shaping model behavior through interactive alignment and user modeling to proactively navigate context gaps, synchronizing machine and human language to control models more reliably, and guiding models to empower humans by teaching us superhuman concepts.

Bio:

Justin Cho earned his PhD at University of Southern California, where he was advised by Jonathan May. His research focuses on building human-centric AI by maximizing shared contextual understanding between humans and AI systems. Prior to USC, he worked at ISI as a programmer analyst and earned his bachelor's degree in computer science at HKUST. He had previously interned at Meta and Amazon AGI. He led the first team from USC to compete in the Alexa Prize Socialbot Grand Challenge and made it to the semifinals. He also organized the Conversational AI workshop at ICML 2023.