This fall, the people who built today's most advanced AI stood up in public and said the quiet part out loud: they're not sure they trust what they've made. Researchers resigned. Executives called for the industry to slow down. Employees put probability estimates on catastrophic outcomes. For clinicians being asked to lean on these same underlying technologies at the bedside, that's not an abstract debate — it's a direct question about the tools already in your workflow. This session sits with that discomfort rather than resolving it: what does it mean to adopt a technology its own creators are publicly hedging on? No tidy answer here — just the question, carried into the panel that follows on what clinicians should still be trained to do themselves.