Watch: Would you trust an AI surgeon?

Autonomous surgery sounds like the kind of thing that belongs in a Michael Crichton novel, not a hospital schedule. But that's exactly the gap this new podcast conversation tries to close, sitting down with Mathias Unberath of Inner Logic to separate what AI-assisted surgical systems can actually do right now from what's still speculative fiction.

Unberath's answer, in short, is that today's systems are nowhere near a scalpel-wielding robot making its own calls mid-operation. Instead, the real progress is happening in narrower, less cinematic places: image guidance, tracking instruments in real time, flagging anomalies a tired surgeon at hour six of a procedure might miss. It's assistance, not autonomy, and the distinction matters more than the marketing around it usually admits.

The conversation also gets into why full autonomy is such a stubborn problem. Surgery isn't a static dataset you can train a model on and call it done. Every patient's anatomy varies, tissue behaves unpredictably, and a system has to perform reliably under conditions where there's no room for a confident-sounding wrong answer. That's a very different bar than generating a plausible paragraph or classifying an image. Get it wrong in a chatbot and someone rolls their eyes. Get it wrong here and someone doesn't wake up the same.

What comes through in the episode is less a promise of near-term robot surgeons and more a case for patience. Inner Logic's work, as described, is about building trust incrementally: proving out narrow capabilities, keeping humans firmly in the loop, and treating reliability as the actual product rather than a footnote to a demo reel. It's a slower story than the sci-fi version, but probably a more honest one.