Contextualization for Human-AI Interactions

This seminar explores how contextual understanding is crucial for effective human-AI interactions. Presented by Justin Cho from USC/ISI on February 27, 2025, the talk addresses the limitations of current AI systems despite their impressive capabilities. Learn about three key research directions that demonstrate how utility depends on context: adapting language models for speech-based interactions beyond text interfaces, personalizing AI responses to individual users with minimal data, and developing holistic evaluation approaches for complex tasks like content moderation. The speaker, a PhD Candidate at USC advised by Jonathan May, brings expertise from leading USC's team in the Alexa Prize Socialbot Challenge, co-organizing the Conversational AI workshop at ICML 2024, and internships at Meta and Amazon.