Latest / AI Ethics with Fexingo: Bias, Safety, and Responsible Artificial Intelligence
When Your AI Therapist Gives Bad Advice Based on Biased Training Data
Lucas and Luna dive into a troubling case in mental health AI: a chatbot therapist trained on a dataset that over-represents certain demographics and therapeutic approaches, leading to harmful advice for users outside that narrow lens. They unpack a 2025 study from Stanford's AI Ethics Lab that found a 34% higher rate of inappropriate responses for non-white, non-English-speaking users, and discuss why this isn't just a 'bias alert' but a question of safety regulation. Lucas explains how the training data — scraped from Reddit's therapy forums — skewed toward a specific cognitive-behavioral…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.