Editor’s Note: An Eye on AI
By Lee DeOrio
For The Record
Vol. 33 No. 3 P. 4
I’m not a medical expert. I’m not a mathematician. I’m a journalist, a group of professionals not normally associated with figuring out algorithms. But algorithms play a significant role in our everyday lives, sometimes blatantly, other times surreptitiously.
For example, algorithms can choose our music. Personally, I take exception when “something” suggests that I’d like a particular artist. If I happen to cave in and give their suggestion a chance, I do my utmost to absolutely despise the poor soul who happened to land inside my headphones.
These algorithms eliminate the work for the listener but, in the process, take the joy out of discovering new artists. How rewarding is it to listen to your favorite independent radio station and have something catch your ear? Next thing you know, you’re on the artist’s Bandcamp page learning more about their work and determining whether that one tune is an aberration or part of a stellar catalogue.
In health care, artificial intelligence (AI) is gaining a stronger foothold, from analyzing large data samples to planning patient care. For example, I can tell by the questions posed to me by my family physician that he’s following a pseudo script in which one answer will lead to another question and so on until he (or the AI) arrives at a conclusion (stop scouring the internet for new bands and exercise more!). Trouble is, the physician doesn’t know that I spend an inordinate amount of time searching for off-the-beaten-path sounds. Let’s see what that does to your “equation.”
On a grander scale, AI extracts data from millions of patients to reach a general conclusion about the ideal path toward better care. No doubt, it’s a powerful tool that the profession can use to get a better grip on diseases and conditions such as cancer and COVID-19. What I’m questioning is how that information scales down to the patient level, where numerous circumstances create a unique treatment setting. Many of the ingredients that make up social determinants of health will throw a wrench into any AI-generated data.
If the objective is to check a few boxes, let the AI work its magic, and send patients on their way, then there’s a good chance the technology will become an abject failure. It may be preferable to detour from a structured, calculated method in favor of a more intuitive approach that takes into account unique circumstances and individuality, both in terms of the patient and the physician.
Otherwise, you may as well just listen to those radio stations whose playlists are dictated by corporate music labels.