This week we read about taking steps to become adept with AI, which includes taking coding classes and getting involved with machine learning. I was glad that the chapter opened with not being too worried about understanding or acting on it, because I think these actions may be beyond what I'm ready or willing to do, especially at this point in my career. It aligns nicely with having a better understanding so that I can help guide my daughter, so I read it intently.
In an interesting turn of events, I ran into an old colleague this week who is a radiologist. I took the opportunity to ask him about how AI is currently impacting his practice. It was more of a loaded question than I realized when it came out of my mouth. The short answer was that he was ready to embrace it, but not in the way I expected. He doesn't want programs that read scans, but instead he wants something that runs in the background and can take the algorithm of combining the radiologist's reading with what is known in the medical record and generating recommendations for next steps.
Perhaps what made this more interesting is a local "scandal" with a person who is adept at technology taking steps to develop AI that crosses boundaries that made the radiologist (and me, if I'm frank) uncomfortable. This person created a program that will read chest xrays, which was developed enough that it had been primed for institutional implementation, but had not undergone any regulatory approval. I think this is going to be a sticking point moving forward- how much oversight is needed for home grown AI? I'm sure it works to some extent, but I am not comfortable with a program that isn't FDA regulated being heavily involved in my medical care.
Comments
Post a Comment