Part of me worries about outsourcing reasoning, analysis, and thought to AIs for things like medical diagnosis

But the bigger part of me wonders if perfect recall and training on every medical text ever might be lead to better outcomes regardless

Trying hard to be the training data I want to see in the world | DΞF | Head of Product Serotonin | building serotonin.co/platform

They're good hesitations. Co:pilots for clinicians and low acquity patient triage are here already. 

It'll be a while before any of it is fully offloaded. Possibly not in my lifetime.