Where Are We Now? Bias in Health AI

Bias in health AI can shape who gets care, how fairly risk is measured, and whether automation helps or harms patients. Karandeep Singh, M.D., M.M.S.C. explains that predictive AI can reflect historical, representation, measurement, learning, evaluation, and deployment bias, especially when models are trained on limited populations or use flawed proxies for illness and access to care. Singh also describes generative AI as a system trained first to predict text and then to follow instructions, with bias entering through training data, instruction tuning, prompts, and outside information sources. Alongside these risks, he highlights practical uses such as AI-assisted sepsis quality review and patient outreach workflows, while emphasizing governance, human oversight, disclosure, and careful measurement of whether these tools actually improve care. Series: "Exploring Ethics" [Health and Medicine] [Humanities] [Science] [Show ID: 41365]

You can listen to Where Are We Now? Bias in Health AI online on Radio and Podcast. Open the player on this page to stream the available audio.

Where Are We Now? Bias in Health AI is an episode from Health and Medicine (Audio) by UCTV.

This episode is 00:35:15 long.

This episode was published on Apr 20, 2026.

Yes. Use the heart button on the episode page to add it to your favorite episodes list.

Yes. This page shows related episodes from Health and Medicine (Audio) when more episodes are available from the podcast feed.