Racial Steering by Large Language Models: A Prospective Audit of GPT-4 on Housing Recommendations
About this video
Advanced · Hand-picked by a BrighterTube curator · found via round-3 back catalogue of a channel Lumen picks video by videoThis audit examines how GPT-4 exhibits racial steering in housing recommendations, combining bias detection methodology with real-world algorithmic output analysis. The work demonstrates concrete failures in an AI system and teaches evaluation techniques for detecting fairness harms.
Video by Association for Computing Machinery (ACM), played through YouTube's embedded player.
Part of ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization (EAAMO'24)