This dynamic fishbowl debate brings engineers, clinicians, and regulatory leaders into a rotating inner‑circle conversation, creating a rare, real‑time exchange between those designing AI systems, those deploying them, and those overseeing their safety. With audience members stepping into the discussion throughout, the session becomes a living model of the very human‑machine interaction it examines.

• Discuss how assistive, decision‑support, and autonomous AI shift risk, validation, and oversight, enabling attendees to match autonomy levels to the appropriate regulatory and safety strategy
• Examine where AI improves consistency and where it introduces over‑reliance or hidden errors, giving participants clearer judgment on when human review adds safety value
• Explore how claims, labelling, and clinical context constrain autonomy, helping attendees align design decisions with regulatory positioning
• Break down models for “human in the loop,” “on the loop,” and supervisory control guiding teams to design workflows, interfaces, and monitoring that meet regulator and clinician expectations