I am not optimistic here. This is the lion’s den - I’m here to see what the predators have in store for us. Back in early residency it exposed me to CNN and real time object detectors and a co-resident and I made a little proof of concept app that detects ICD/PM on CXR.
So I think there’s value to me here - just no value in proselytizing and apologetics. They don’t like us and are here to eliminate us.
I don’t view it as adversarial. I’ve had productive conversations on HN including some which gave me NLP approaches I hadn’t considered.
Engineers (including myself when I had a health-tech startup before my MD) tend to misunderstand the problem space (simply put they consider radiology a classification task and assume ground truth labels exist/are even possible like for object recognition) so it seems easy to them, but I don’t believe the intent is nefarious.
Perhaps it’s my naivety but I think most smart people at least partly care at improving society on some level (even if they want to make a lot of money doing it) and the physician-services budget is a large line-item that seems like an easy target rather than the ??? to improve inefficiencies and outcomes.
Without a doubt the AI-enabled radiologist will render the non-AI rad obsolete but it won’t replace our profession. I don’t believe anyone with the skillset and experience in the relevant AI tech believes that it will eliminate radiologists (other than maybe Hinton), certainly isn’t the attitude of the pure CS supervisors I’ve had in my training (or the folks I collaborate with now).
So I think there’s value to me here - just no value in proselytizing and apologetics. They don’t like us and are here to eliminate us.