Latest / AI Ethics with Fexingo: Bias, Safety, and Responsible Artificial Intelligence
When Your AI Recruiter Questions Your Name
Lucas and Luna investigate a 2025 academic study that found AI recruitment tools penalizing candidates with non-Western names — even when identical resumes were submitted. They walk through the controlled experiment, the statistical gap (a 14 percent callback difference for identical qualifications), and why fixing the training data isn't as simple as stripping out names. They also discuss the technical root cause: how an AI's 'residual name embeddings' still leak demographic signals after explicit removal, and why recent European AI Act provisions on bias auditing may force deeper changes…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.