GPT-6 Astra just solved a problem medical students have been facing for decades Medical students have been working around the same three problems for decades. Someone talked to them, identified all three, and built a service with Astra that solves them in one tool. The problems are structural. Flat anatomical diagrams strip depth out of organs entirely. CT cross-sections don't connect naturally to the overview diagrams students spend years memorizing. And reading a scan requires holding three mental models simultaneously - patient orientation, cross-section direction, and how every organ shape shifts through each layer. Astra built a service that handles all three in one place. Not a better textbook. A spatial tool that connects the diagram, the CT view, and the 3D relationship together. Medical education hasn't changed its core format in over a century. This might be the first tool designed around how anatomy is actually understood - not just displayed.

Is always-on ambient AI - i.e. AI that hears and sees everything - the "holy grail" of healthcare? Or a dystopian surveillance state that will scare off human clinicians altogether? Recently a family member saw a doctor. No AI scribe - just a good ol' human clinician doing normal medicine like a sane person 😉. After, we checked the patient portal and noticed things in the note that did NOT actually happen: → "1 week of symptoms"... told them it was a 2 day history → "Surgical history reviewed - none pertinent"... nope, was never asked → "Pupils are equal, round and reactive to light"... wait, when did an eye exam happen? Did any of this materially impact the diagnosis or management of the patient? No. Would any of this staying in the record actually matter? Probably not… unless "no surgical history" got copied/pasted and people stopped asking. I suspect this clinician went on auto-pilot with smart phrases, needing boxes checked for "compliance" and "documentation." The most charitable read is that they genuinely misremembered. You'd think AI scribes fix this, but they don’t completely - not just yet. Because even in the age of AI scribes, humans are still in the loop - they edit the note and "verify" the AI output. Humans still "verbalize" the physical exam so the AI scribe even knows one happened. Which means if a clinician skipped something needed for compliance, they could just say they did it - and we just moved the loophole from paper to software. The fix, in theory, is visual ambient - cameras everywhere with computer vision AI. One day we won't need to tell the AI scribe we did an exam - it'll just watch us do it. Yet what happens if the AI scribe sees we skipped something, hears we didn't ask something - but the clinician says "yes I did"? What ethical responsibility does AI have to monitor and report what humans do? Should we care? On one hand, I dislike the idea of a surveillance state in healthcare - cameras and microphones recording all the time gives me the creeps. On the other, there's the bull case for always-on Ambient AI. We'll catch falls and deterioration faster, and we’ll measure vitals without extra hardware. It'll be a huge deterrent for bad actors - not the people bad at documentation, but clinicians who truly harm patients. Always-on AI will weed out these bad humans faster. But amid all that upside, this surveillance and documentation baggage comes with it - whether we like it or not. Maybe none of this will matter. The AI is always there on a Star Trek ship - and no one there seemed spooked by it. Most likely it'll be normalized and clinicians will be fine with it… but should we? And if the price of finally knowing whether someone truly did a physical exam is Big Brother standing over us in the room - is that a price we're willing to pay? In the era of AI, we’ll quickly stop asking whether something is possible - and think more about whether it should.