From a radiology world this is fascinating. I'm not worried about job security as I'm an interventionalist. What I'm wondering is about go-to-market strategies for diagnostics. I do some diagnostic reads and I would love to have something like this pre-draft reports (especially for X-Rays). There are tons of "AI in rads" companies right now, none of which have models that come anywhere close to GPT-4 or even this. Personally, I would not start such a company because the barrier to entry is now so low. I think there needs to be an open-source project that just subtrains these models on rads images and is available to all radiologists.
