About this video
Keshavan Seshadri joins Chris Hutchins to explain why general-purpose intelligence is unsafe in healthcare without patient, task, operational, and institutional context. They examine confidence scoring, explainability, bias detection, guardrails, and when an AI system must defer to a human.
Topics covered
- The four types of context a healthcare AI system needs
- Why confidence scores and risk determine human deferral
- How explainability, logging, and tool tracing support accountability
- Why regulated AI needs continuous evaluation and guardrails
This conversation is educational and does not constitute medical, legal, or regulatory advice.
Featured guest
Keshavan Seshadri
Senior ML Engineer
Machine-learning engineer and AI platform builder focused on context-aware systems, explainability, evaluation, and human-in-the-loop design.
More conversations like this
Stay in the signal
Subscribe for new conversations on healthcare AI, leadership, governance, and trust.
Subscribe on YouTubeHelp others find the show
Loved this episode?
A 30-second Apple Podcasts review helps more healthcare AI leaders discover The Signal Room.
Leave a Review