Many clinicians are skeptical about AI in healthcare. But some are ready for the disruption

By now, you’ve been hearing a lot about ChatGPT. And you’ve probably had a gut reaction to it – maybe not a good one.

But what if artificial intelligence could help doctors and other healthcare providers with some of the grunt work of practicing? Like answering non-clinical questions in MyChart messages or creating after-visit summaries. Or helping patients with appointment scheduling, freeing up front-office staff to focus on other tasks.

Those are the sorts of areas that excite John Whyte , the chief medical officer of WebMD . ChatGPT, a tool from for-profit lab OpenAI , is essentially a chatbot that mimics human communication. The concept isn’t new – chatbots have been used for quite some time across a range of industries – but the buzz these days is around how good the technology has become for long-form writing. ChatGPT can draft term papers and legal documents. It has enough creativity to produce advertising copy. It can pass parts of the U.S. Medical Licensing Exam.?

The Journal of the American Medical Association even published an editorial this week updating its guidelines around the use of AI in manuscripts and published articles.?

Whyte acknowledges that many of his fellow physicians are skeptical about the technology. The stakes are particularly high in medicine when it comes to misinformation, and tech companies have been accused of not being transparent enough about the underlying algorithms being used. Whyte also spent nearly five years at the Food and Drug Administration, and it’s not lost on him that none of these tools are regulated – even as some healthcare startups (particularly in mental health) have used chatbots in ways that have raised concerns about how well the technology can respond to people in crisis.

But Whyte thinks AI is only going to get better, and will be something physicians will increasingly need to understand and adopt. And maybe even come to appreciate?

What do you think? What are the opportunities – and drawbacks – for ChatGPT in healthcare?

The following transcript of our conversation has been lightly edited for length and clarity.

LinkedIn News: What do you think are some of the use cases for ChatGPT in healthcare?

Whyte: I have young kids. It's impossible to get ahold of the pediatrician; their online scheduling system is horrible. It's not productive, and that's a lot of the healthcare system. But if we could have a natural language processing chatbot that can help automate in a better way, that alone would help.

I still see patients. I get a lot of emails from patients, and I get a gazillion that are like, what are my lab results? Not: Can you explain my lab results? But: what are my results? If you can message me on the portal, you actually can see your labs. That's not the best use of my time, and we're not communicating to patients well enough. I think there is a lot of opportunity there.

LinkedIn News: Other industries already use chatbots for their customer service.

Whyte: There is a consistency in terms of what people ask – there really is. I heard from a colleague the other day who said, It would be great if ChatGPT could do insurance denials; responses to that take a lot of work. There are a lot of ways it could free up time.

LinkedIn News: Where do you think the objections are coming from?

Whyte: None of these AI products right now can really simulate empathy. There's still an art to the diagnosis in terms of the history and the physical. That's not going to go away. But can it actually help me make a diagnosis quicker? Could it summarize data for me from numerous studies that I don’t have time to read??

I think the issue for a lot of my colleagues is the issue of trust. They don't trust it. They don't know how it works. They don't know how it gets this information. They want to know that they're comfortable with that decision-making in front of the patient. And then it's also that we've been told so many times that something new is going to make our lives easier and – initially at least – it doesn't.

Physicians tend to be a cynical group. They don't tend to be the early adopters of technologies. But I think it'll remove a lot of the tasks that we do that really don't seem very doctorly.?

I also think it'd be great if it could [improve upon] diabetes education or heart disease prevention. A lot of that is lifestyle: you need to lose weight, you need to eat healthy, you need to eat fruit and vegetables. Those aren’t personalized recommendations. Could ChatGPT do that?

LinkedIn News: Mental health apps have tried to do that, but they’ve missed cases of suicidality or responded inappropriately.?

Whyte: Chatbots ended up being terrible when giving that advice. That's absolutely right. There's very mixed data on that. But that shows the empathy component, the art of medicine.?

I'm looking at it as [a way to] help more with triage. It can provide personalized information and recommendations to people. For example: I’m overweight and over 65; how can I start an exercise program? Or: Give me five examples of exercise. Doctors aren't good at that. They're not interested in doing that. And it's hard for patients to comb through the internet and go through all that.?

LinkedIn News: Isn’t that essentially what you’re doing at WebMD??

Whyte: If you search for symptoms of [sexually-transmitted diseases], we can send you a link to get a lab test without seeing a doctor, if that's allowed in your state. It’s recognizing that people don't just want to search content anymore, and then wait six weeks to see your doctor.?

I think it's the same thing with a lot of these apps. I do think there needs to be more regulation. But [artificial intelligence is] only going to get better. That's why I think people have to start to embrace this. It's not going to go away. It's going to be iterative.