AI for Email Replies? Not Yet

An article in the March 20 JAMA Network Open looked into the use of AI for responding to patient emails. Basically, they found that this led to a reduction in physician burden, but didn’t save any time.

1. Not sure that’s worth the trouble.

2. Unless the AI is simply responding with something like “message received, thank you” I don’t think this is a good idea.

Yeah, we’re all stretched for time, I understand that. From the starting gun each morning we’re racing between patients, phone calls, incoming test results, staff questions, drug reps, sample closets, dictations, and a million other things.

But AI needs to be a lot better before I let it handle patient questions and concerns.

Someday, yeah, maybe it can do this, like 2-1B, the surgical droid that replaced Luke’s hand in “The Empire Strikes Back.” But we’re not even close to that. Just because a log-in screen says “Jumping to Hyperspace” doesn’t mean you’re on the Millennium Falcon.

I generally know my patients, but even if I don’t remember them, I can quickly look up their charts and decide how to answer. AI can look up charts, too, but data is only a part of medicine.

There are a lot of things that don’t make it into a chart: our impressions of people and a knowledge of their personalities and anxieties. We take these into account when responding to their questions. People are different in how things need to be said to them, even if the answer is, overall, the same.

“It’s the AI’s fault” isn’t going to stand up in court, either.

I also have to question the benefit of the findings. If it lessens the “click burden” but still takes the same amount of time, are we really gaining anything?

I’m all for the digital age. In many ways it’s made my practice a lot easier. But I think it has a way to go before I let it start dealing directly with patients.

Dr. Block has a solo neurology practice in Scottsdale, Arizona.

An article in the March 20 JAMA Network Open looked into the use of AI for responding to patient emails. Basically, they found that this led to a reduction in physician burden, but didn’t save any time.

1. Not sure that’s worth the trouble.

2. Unless the AI is simply responding with something like “message received, thank you” I don’t think this is a good idea.

Yeah, we’re all stretched for time, I understand that. From the starting gun each morning we’re racing between patients, phone calls, incoming test results, staff questions, drug reps, sample closets, dictations, and a million other things.

But AI needs to be a lot better before I let it handle patient questions and concerns.

Someday, yeah, maybe it can do this, like 2-1B, the surgical droid that replaced Luke’s hand in “The Empire Strikes Back.” But we’re not even close to that. Just because a log-in screen says “Jumping to Hyperspace” doesn’t mean you’re on the Millennium Falcon.

I generally know my patients, but even if I don’t remember them, I can quickly look up their charts and decide how to answer. AI can look up charts, too, but data is only a part of medicine.

There are a lot of things that don’t make it into a chart: our impressions of people and a knowledge of their personalities and anxieties. We take these into account when responding to their questions. People are different in how things need to be said to them, even if the answer is, overall, the same.

“It’s the AI’s fault” isn’t going to stand up in court, either.

I also have to question the benefit of the findings. If it lessens the “click burden” but still takes the same amount of time, are we really gaining anything?

I’m all for the digital age. In many ways it’s made my practice a lot easier. But I think it has a way to go before I let it start dealing directly with patients.

Dr. Block has a solo neurology practice in Scottsdale, Arizona.

An article in the March 20 JAMA Network Open looked into the use of AI for responding to patient emails. Basically, they found that this led to a reduction in physician burden, but didn’t save any time.

1. Not sure that’s worth the trouble.

2. Unless the AI is simply responding with something like “message received, thank you” I don’t think this is a good idea.

Yeah, we’re all stretched for time, I understand that. From the starting gun each morning we’re racing between patients, phone calls, incoming test results, staff questions, drug reps, sample closets, dictations, and a million other things.

But AI needs to be a lot better before I let it handle patient questions and concerns.

Someday, yeah, maybe it can do this, like 2-1B, the surgical droid that replaced Luke’s hand in “The Empire Strikes Back.” But we’re not even close to that. Just because a log-in screen says “Jumping to Hyperspace” doesn’t mean you’re on the Millennium Falcon.

I generally know my patients, but even if I don’t remember them, I can quickly look up their charts and decide how to answer. AI can look up charts, too, but data is only a part of medicine.

There are a lot of things that don’t make it into a chart: our impressions of people and a knowledge of their personalities and anxieties. We take these into account when responding to their questions. People are different in how things need to be said to them, even if the answer is, overall, the same.

“It’s the AI’s fault” isn’t going to stand up in court, either.

I also have to question the benefit of the findings. If it lessens the “click burden” but still takes the same amount of time, are we really gaining anything?

I’m all for the digital age. In many ways it’s made my practice a lot easier. But I think it has a way to go before I let it start dealing directly with patients.

Dr. Block has a solo neurology practice in Scottsdale, Arizona.