Human-AI Interaction: From Supporting Surgical Training to Inspecting Social Bias in LLMs

Explore cutting-edge research in Human-AI interaction through this informative talk by Rafał Kocielnik, a Postdoctoral Researcher at Caltech. Delve into two distinct projects showcasing opportunities and challenges in modern AI applications. Learn about leveraging multimodal deep-learning techniques to analyze formative feedback in surgical training, a project that received the best paper award at the ML4H conference. Discover the BiasTestGPT framework, designed to empower domain experts in inspecting generative AI for harmful stereotypes. Gain insights into generating high-quality synthetic data for social bias testing at scale and explore a user-friendly interface for public engagement in AI inspection. Benefit from Kocielnik's interdisciplinary expertise spanning computer science, industrial design, and human-centered engineering as he discusses the intersection of AI and HCI in specialized domains like surgical education and gaming.

Human-AI Interaction: From Supporting Surgical Training to Inspecting Social Bias in LLMs