🚨 AI IS PUTTING DOCTORS TO SHAME

A Harvard study published this week found that AI outperformed emergency room physicians at diagnosis across multiple categories of case complexity.
67% accuracy on triage cases. The two attending physicians hit 55% and 50%.
The lead researcher's quote: "We tested the AI model against virtually every benchmark, and it eclipsed both prior models and our physician baselines."
Not "it's getting close." Not "by 2030." Already at the ceiling.
OpenAI's GPT-5.5 launch last week showed the biggest gains specifically in early scientific research, areas that are, in their words, "very intelligence bottlenecked."
Drug discovery. Diagnostics. Protein folding. Clinical trial design. Every one of these fields runs on intelligence as the primary constraint.
NVIDIA's new infrastructure is cutting the cost of running frontier AI by 35x per token. The economics of deploying AI in healthcare, historically prohibitive, are about to change faster than most hospital systems are prepared for.
Biotech was always going to be the next AI. Not because biology is easy, but because AI is making biology tractable at a speed no human research team can match.
The capital is starting to figure this out. The policy hasn't caught up yet. That gap is usually where the opportunity lives.