I would be comfortable with CaBOT being an adjunct, and even more comfortable if it was blinded. My concern would be it becoming a crutch. I don't think we sho

I would be comfortable with CaBOT being an adjunct, and even more comfortable if it was blinded. My concern would be it becoming a crutch. I don't think we should allow general skepticism in AI (which I share btw) to blind us to specific use cases where RAG/LLM are doing things which can be shown to be as good as or better than an expert in role, at least the expert you can afford to see. One downside of improved imaging has been an increase in "detected marker for..." causing patients to present. Not all of these need to. It takes time to learn how to nuance improved diagnostics and I think CaBOT will be part of that on both the good and the bad side. "It's always Rhabdo" (mandatory House quote) (not a doctor)