This TIME Lectures episode presents a morbidity and mortality style examination of three Australian coronial cases involving sepsis, post operative deterioration, and inpatient falls, asking a deliberately narrow question: where could AI embedded in hospital workflows have changed the outcome. Anchored closely to coronial findings, the discussion avoids speculation and focuses instead on system level failure modes such as delayed recognition of deterioration, fragmented data synthesis, discretionary escalation, and loss of trajectory across time and handover. Rather than positioning AI as a diagnostic replacement, the episode explores its potential role as a reliability layer, enforcing escalation, integrating existing clinical data, and reducing preventable variation in care. Designed for senior clinicians and health system leaders, the session treats AI not as a solution in search of a problem, but as a possible response to recurrent patterns seen in contemporary morbidity and mortality review.