Aaron Lee, MD, an associate professor of ophthalmology at the University of Washington, discusses some limitations of artificial intelligence (AI) in the ophthalmology field.
Artificial intelligence (AI) can't do the full spectrum of all the things a human clinician does today, said Aaron Lee, MD, an associate professor of ophthalmology at the University of Washington.
Can you discuss some of the pitfalls resulting from the transition to digital care in ophthalmology?
I do worry a little bit about the hype around AI, because so much of AI has elevated everybody's expectations. We expect AI to be able to drive our cars. We expect AI to keep airplanes in the air and all those kinds of things. And it is true that this technology, specifically deep learning, is a remarkable advancement in artificial intelligence. It is true that it is capable of some wondrous things, things that I thought were completely impossible before. But it comes with very important caveats.
They're very, very good at a very narrow window of things. So I think people may not understand, both clinicians as well as patients, that an AI algorithm being used in their care may only be useful in that very narrow context. Now, in that context, it may do an amazing job, but it can't do the full spectrum of all the things that a human clinician can do today.