Using a scene from the TV show The Pitt as a lens, this piece argues that the real barrier to clinical AI adoption isn't model accuracy — it's liability and institutional economics. When an AI tool flags or documents something, hospitals become legally responsible for acting on it, creating a searchable trail of omissions. Compounding this is a structural trap: AI needs to learn from doctors' clinical reasoning, but that reasoning is deliberately kept out of medical records to avoid legal exposure. The post advises buyers to ask who bears risk when the tool is wrong, and builders to design for physician judgment to remain in the loop rather than optimizing purely for detection accuracy.
Table of contents
The wall isn’t intelligence.Detection creates a duty.There’s a trap underneath the wall.Here’s what this means if you’re building or buying.Back to The Pitt288 Impressions