Atmosphere TV is the largest streaming TV platform purpose-built for shared, social spaces like bars, restaurants, gyms, and airports. Atmosphere TV is the creator of ChiveTV and 30+ mesmerizing, audio-optional channels that entertain over 150 million monthly viewers.

Built for business, made for fun. Atmosphere TV. Learn more.

Marcus Johnson: [00:00:00] Atmosphere TV is the largest streaming TV platform purpose-built for shared social spaces like bars, restaurants, gyms, and airports. Atmosphere TV is the creator of CHIVE TV and over 30 mesmerizing audio optional channels that entertain over 150 million monthly viewers. Built for business, made for fun, Atmosphere TV.

Hey, gang. It's Monday, October 5th. Rajeev, Beth, and listeners, welcome to Behind the Numbers: New Market podcast, made possible by Atmosphere TV. I'm Marcus. Joining me for today's conversation we have two of our health and pharma senior analysts living in New Jersey, Rajeev Leventhal.

Rajiv Leventhal: Hey, Marcus. How are you?

Thanks for having

Marcus Johnson: me. Hello. Very good, mate. Good to see you. Thanks for being here. We're also joined, um, in Pennsylvania by Beth Snider Buelick.

Beth Snyder Bulik: Hey, Marcus. Nice to see you. Hi.

Marcus Johnson: Good to see you, too. Today's fact.

Dr. James Naismith, who invented basketball, is the only Kansas men's [00:01:00] basketball head coach with an overall losing record. That's tough. Hmm. Although, it sounds like the England men's soccer team. We invented the bloody game. We can't win a major tournament. But I feel his

Beth Snyder Bulik: pain. Anyway- What is that saying?

Those who can, do. Those who can't, invent?

Marcus Johnson: Probably.

Rajiv Leventhal: I wonder what the big obstacle was. Wow.

Marcus Johnson: For England? How long do you have? For Naismith, no idea. Anyway, today's real topic, how people use AI for health.

All right, gang, two years ago we discussed on this show what AI should and shouldn't be allowed to do in healthcare. Uh, Rajiv, you've been looking into this recently. What's the main way, in your opinion, people are using AI in healthcare differently? What's changed since then?

Rajiv Leventhal: Yeah, so I think the main way is still the same [00:02:00] as it was a year or two ago.

People are using AI as a health information tool, right? To look up, uh, information about symptoms or their he- a health issue, or to, you know, kinda have, like, a conversation based on their personalized situation. Like, "Here's what I'm feeling, here are my symptoms." Uh, what do you think I should do? Things like that.

That really hasn't changed, but what has, in my research, I found that AI is moving from st- a health information tool, but to a really a decision-making layer and, and a, a tool that is actually changing healthcare decisions. So beyond asking AI what a symptom means or researching a condition, people are increasingly using its guidance to decide whether to seek care, uh, even to, you know, shop for a healthcare product, like an over-the-counter health product, um, and what steps to take next.

So I think that's been the, the most, uh, important acceleration of health [00:03:00] AI activity in the last year or so.

Marcus Johnson: Yeah. You have some fascinating numbers to support that. One from Rock Health Survey, uh, 2025, December, 81% of people among- amongst AI users who asked a chatbot a health question, 81% acted on the result.

Uh, another stat from eHealth Survey, uh, from May of this year, 73% of people who use AI for health advice changed their care. Um, and also a second opinion, so people, um, close to 70% of patients have used AI after a doctor's visit, according to a September survey from this year, um, whether they're trying to, you know, evaluate treatment options, um, uh, figure out, um, uh, what, what the test results are telling them.

Um, so that- Yeah ... big change of they're still looking, uh, maybe more of them arguably looking up information about health, more people acting on that information.

Rajiv Leventhal: Yeah, and I think sometimes the natural assumption is that, uh, if AI is influencing decision-making, it's a negative. [00:04:00] But that's not, uh, necessarily the case, right?

Sometimes, you know, people's, uh, tendency is to sort of, uh, you know, you know, overact and,

Marcus Johnson: and maybe, uh, because of anxiety and,

Rajiv Leventhal: and fear of what's going on, they go to the emergency room when they don't have to, right? Mm-hmm. So it's not always the case that AI is telling you not to go to the doctor when you should be.

There's lots of instances where AI becomes, like, the first point of a healthcare journey where you then... it then takes you to other sources to look up information. It takes you to a doctor or maybe avoids an unnecessary, a trip to the doctor. So there are ways where I think, um, AI as a decision-making tool- Mm

you know, has a positive impact. Really- In addition to, of course, the ways it might have a negative impact.

Marcus Johnson: Yeah, well, really quickly, Beth, before I ask you about what, what you think's changed in, in this space, um, 44% of folks have little or no confidence in their ability to discern true health information, uh, nearly half according to a May 2026 KFF poll.

So Rajeev, how can consumers distinguish good [00:05:00] AI-generated health advice from the bad?

Rajiv Leventhal: Yeah, I mean, that's a, a really important question. Um, I think that- It really comes down to the, like, like AI generally, right? It's, you know, information in affects information out. So, uh, I cited a, a study in the, i- in my report looking at, um, how AI fared when it was given, you know, kinda context and patient information and clinical information, and a lot of the times when patients didn't get the most reliable, trustworthy output based on, you know, how it was assessed by, by clinical researchers, it was because the prompt was missing really important contextual details, right?

Mm-hmm. So AI performs a lot better when it has this complete clinical picture than when it receives limited kinda patient-written descriptions. But a lot of people, you know, obviously AI being used for this case is very new, right? So a lot of people, uh, y- and, and they don't have the medical expertise, so they're not quite [00:06:00] sure what to input.

Right. And that, I think, is generally, uh, when sometimes, like, the bad AI health advice happens.

Marcus Johnson: Mm-hmm.

Beth Snyder Bulik: Right. If they're using it like they use a search engine and just asking general questions versus giving precise, you know, good prompts get good answers.

Marcus Johnson: Yeah.

Beth Snyder Bulik: And that ha- the same thing happens in, in healthcare.

Marcus Johnson: Yeah, I mean, r- really good answers. Th- there were some numbers you were showing that, um, it's, you know, okay, uh, at figuring out, you know, the, the, the, the right next step if you give it some information. If it has a complete picture, it is right a lot of the time, um, which I thought was, was fascinating. Beth, what, uh, uh, s- taking a step back again to the original question, what are the main ways that you think people are using, uh, AI in healthcare differently today?

Beth Snyder Bulik: One of the things that people are doing is taking their information that they find on AI to their doctors, um, and asking them, like, "I found this." Again, it's a little bit of the sort of search, what you find on search, um, [00:07:00] about healthcare, I mean, about your, um, health condition. You take to your doctor and ask them about that.

But now they're taking AI in- AI-generated information, um, and saying, "Okay, this is what I found out. What do you think?" I think doctors, um, most doctors say, right, more than half your doctors say that patients frequently or very frequently bring AI-generated information into appointments.

Marcus Johnson: Mm-hmm. Not always received well, correct?

Beth Snyder Bulik: Sure. I mean, you know, doctors don't like to be second-guessed, but, um, I do think that we- there- it- they have an openness to it as well. I mean, they want informed patients. They have a limited amount of time to spend with them, and so people who come in at least with some discussion questions, maybe not demanding a prescription, but are asking questions, would be received more-

Marcus Johnson: Yeah

Beth Snyder Bulik: favorably.

Marcus Johnson: Yeah, you can't, you almost can't blame folks for, for doing this. There was a really interesting piece by Pratik Pawar of Vox saying the average ... And we've talked about this before, the average wait for a primary care appointment in the US is now one month, and in some cities [00:08:00] like Boston, it's over two.

When you do get in, the visit lasts about 18 minutes. Uh, and they were saying, they were citing OpenAI, uh, as saying that seven out of 10 health-related ChatGPT conversation's happening when clinics would be closed. Uh, and so because appointments are hard to come by, and also very quick, um, yeah, trying to do a bit more homework or bringing that homework to the appointment, it does, it does make, seem to make a lot of sense.

Beth Snyder Bulik: I think it works on the other end too, what, um, Rajiv said earlier about being helpful. People who go through, you know, you have- Mm ... 15 minutes, 20 minutes with a doctor, and you leave, and you don't really remember or know what they said, AI can be helpful in that case to explain, re-explain, I guess, or go into more detail about what your diagnosis or condition is.

Marcus Johnson: Yeah. Rajiv, they're bringing it to the party to try and help the doctor out or to ask questions about what they found. How much are consumers using AI to stand in, uh, almost as a replacement for traditional healthcare?

Rajiv Leventhal: Yeah, so I tried to look at kind of different categories of how [00:09:00] consumers are using AI for health.

One is clearly what we discussed at the top, a research companion. Another is maybe to supplement care, but not necessarily replace it. So as Beth mentioned, I just went to the doctor's office. I come home. Uh, now I need to kind of have AI maybe interpret some of what was said at the doctor's office. I didn't have more than 20 minutes to, uh, uh, to kind of ask doctor f- follow-up questions, right?

You don't always have a lot of time. Or you think of a follow-up question on the way home from your appointment. Yeah. Well, maybe the doctor isn't accessible right away, so you go to the, to the AI tool. So that's more of like a, what I would say is a, is a supplement or an additional com- complimentary tool.

But then there is, uh, the, the use case of using AI to directly replace, uh, a doctor's advice, which is happening. Uh, our report, uh, referenced a KFF survey which found that among consumers who use AI for health in the past year, almost half, so 46%, [00:10:00] cited unaffordable doctor visits as a reason, and similar percentage, 44%, noted, uh, difficulty reaching a provider.

So that's, uh, according to a March 2026 KFF survey. So for those cases, uh, it's clear that people who are using AI for health ha- you know, they're having trouble reaching a doctor, whether due to a wait time or some other reason, or affording a doctor's visit.

Marcus Johnson: AI is always available. It's around the clock.

It's mostly free.

Rajiv Leventhal: So, uh, you know, that's when it's serving as a stand-in.

Marcus Johnson: Yeah.

Beth Snyder Bulik: Yeah, I think we've talked about the mental health before as well in other conversations, but young people are using it for mental health support, and they're among the, the least, uh, the population that least often goes to see a provider.

Marcus Johnson: Mm-hmm. So people using- Uh, in some instances, uh, uh, AI as a way just to get access to, to, to a doctor, um, whether that's 'cause it's more convenient or because it's, uh, more affordable. [00:11:00] Um, so they're using AI for their healthcare, um, uh, questions, queries. R- Rajeev, how much are consumers turning to AI for health, uh, compared to other traditional online sources?

Rajiv Leventhal: Yeah. So even as more and more people use AI for health, it's still not really close to the top online health information source. Our forecasts project that AI this year will actually be the fastest growing use case among, uh, all AI functions that we survey, or that we study, sorry. Uh, but still, y- uh, I think our, um, our survey found, our, our January survey found that, uh, only about 15.6% of consumers who look up health information online use AI tools.

So that might seem like, you know, still a relatively small, uh, segment of the population even as it does quickly gain traction. Um, and I think people kind of wrongly assume that AI for health is just, like, [00:12:00] used as, like, a replacement for Google in the consumer health information journey, but that's not really the case.

Most people are, who use AI for health turn to Google first or after, um, they turn to Google more often than chatbots, and then other channels are used more often too, like health information websites, uh, social media. They actually outpace AI tools by some distance according to the January survey I referenced.

And, uh, you know, the takeaway there is that consumers are just consulting multiple sources, right? So AI is part of a, part of a broader information ecosystem rather than, uh, a one-for-one replacement for, uh, more established online sources.

Marcus Johnson: Mm-hmm. Um, Beth, how much are folks turning to, um, or- These consumer health products.

Uh, a lot of AI players have launched their own. There's ChatGPT Health, there's- Mm-hmm ... Copilot Health, there's others. Um, w- how are folk- how are they, uh, doing, and, uh, could you see one more than others perhaps gaining traction [00:13:00] amongst consumers?

Beth Snyder Bulik: So, I mean, I think most people, uh, our, our studies, research shows that most people turn to ChatGPT first to ask those health questions.

Mm-hmm. Um, but some of these specialized tools, I guess it's gonna depend on the environment, it's gonna depend on where people are, if you're w- whatever tool you're using. But there's a potential for those specialized tools to be able to give better answers, more detailed. And the p- part of the thing about AI or the thing that AI is adding to these search processes, like Rajeev was talking about, all the different ways people source, um, health information, it's adding a real personalized, so you can add- Mm-hmm

your own personal records, you can put up. So I guess maybe it will depend on w- what engine or what health engine or chat engine, chat bot that people trust. Yeah. So they're going to give it the most, their most personal data, or even upload their medical record, that's possible.

Marcus Johnson: Yeah. Rajeev, well, really quickly on that, I, I thought it was fascinating, this is from Pew Research Center, [00:14:00] um, people willing to share personal health information with chat bots.

Uh, 29% said extremely or very comfortable sharing personal health information, 42% saying somewhat. So that leaves just 26% saying not to, uh, not, uh, not so much or not at all. Uh, Rajeev, for, for folks who aren't too familiar with, with ChatGPT Health, um, as Beth said, over a hund- uh, over 300 million people using ChatGPT each week for health questions, according to OpenAI, who makes it.

Um, what, um, what's the difference in terms of people using ChatGPT for health stuff versus ChatGPT Health?

Rajiv Leventhal: Yeah, I think if you're an AI user, uh, health information and health-focused queries are, have to be one of the top use cases, right? And I think the data, um, backs that up. Uh, because it gives you, as, as we've discussed, it gives you more personalized, uh, responses and guidance [00:15:00] than really you can get on any other health information tool, right?

Like, if you have a, a health issue and you're looking for more information online, and you go to a health website like a Healthline or WebMD, or you just type it into search, although now search is essentially an AI tool on its own, but let's just put that aside for a second. Um, you're just not gonna get the level of contextualized personal details that you will than if you ask the question to an AI, uh, chatbot that, you know, allows of course for like follow-up information and, and contextual details and all of that.

So- Mm-hmm ... uh, yeah, I think that that, for that reason, OpenAI and Microsoft and Perplexity, Perplexity and Anthropic all release consumer health AI tools because they knew how many people were typing in health queries to their, uh, you know, to their leading AI products. Mm-hmm. So they're all kind of competing to s- to kind of keep consumers sticky to their platform.

You know, we can provide the most personalized, customized, [00:16:00] uh, AI health experience answering all of your questions. You can, you know, upload your Apple Health data and your medical records and you, if you want, and your lab history and all this to make sure it's the most, uh, comprehensive, um, individualized answer that we can give you.

But the thing is, like they all, they're all kind of doing the same thing now, so I'm not really sure where the differentiation is. Yeah. But you can see why all of these AI giants have, have turned their, um, tools into consumer, consumer health products. I, I still think there's gonna be a pretty big divide of people who are comfortable sharing, like their really, uh, sensitive like personal- Mm

medical records with, uh, uh, with AI tools and, uh, I think most people will still have their reservations about that, at least for now.

Marcus Johnson: Yeah. Yeah, it'd be interesting to see a list of, uh, types of data people are willing and not willing, uh, or perhaps willing to share. Um, is that-

Rajiv Leventhal: You know, there, there's, there's stories, though.

I, I reference some of this in my report. There's stories of [00:17:00] people who have, you know, been kind of perplexed with, uh, their own diag- lack of diagnoses or their kids, and they've been to a bunch of different specialists and doctors, and you know how the healthcare system works. You just can't get answers.

And they, they put in all of, like, their medical imaging, right? So obviously, you know, sensitive medical details- Mm-hmm ... into, into, uh, AI platforms, and it gives them the answer that the, the human doctor couldn't. So there's definitely, uh, plenty of cases out there that shows, you know, AI's pretty impressive capabilities.

Marcus Johnson: Mm-hmm. One of the things I thought was most impressive about it, Ina Fried of Axios noting Microsoft said, "Copilot Health will let users combine medical records," so lab results and wearable data. So you could take stuff from Apple Health, from Oura Ring, from Fitbit, uh, and put that all into the, to the same place.

Y- if it's gonna be personalized, you're gonna need the, the data to, to do so. Um, Gina Kolata of The New York Times w- writing a piece titled AI Is Making Doctors Answer a Question: [00:18:00] What Are They Really Good For? She cites Dr. Jonathan Chen, an internist at Stanford, who says, "AI programs are becoming existentially threatening for doctors.

They threaten your identity and your purpose." Um, Beth, what is the role of the doctor in an AI world?

Beth Snyder Bulik: I mean, I think the, we, every, our s- our surveys, uh, research shows people trust their doctors. So it is, maybe it's validation, but it's definitely that trusted, um, in-person or however you're t- speaking to your doctor, but that sort of trusted medical, real, y- you know, where does the AI come from?

You definitely, you get the AI result, you look at, you trust it, but you still go to your doctor. I mean, that's, that's the role, the, the doctor. Maybe, maybe some of the diagnostic turns over to AI. Mm. But for the, for that trusted relationship, do you have a relationship with AI? I mean, I guess we can debate that.

Some people with their mental health do have relationships or use them more [00:19:00] closely. But I think in, in general, you want that validation or that- Mm ... medical opinion from your doctor. And that, and like I said, studies show that doc- people trust their doctors more than anything. Well, we

Marcus Johnson: start to see that less and less though, Rajiv, over time, especially as people can't get to their doctor as much.

Maybe, you know, um, there's a shortage of doctors, so a lot of the time you're not getting in to see a doctor, you're getting in to see, uh, another type of physician. Uh, there was this stat which really jumped out me, at me from the Temra survey. 17% of people followed AI advice even if it conflicted with their physician's guidance.

Rajiv Leventhal: Right. I think it's a little early, but I think, I think it's gonna be interesting to, to kinda follow and track the impact that AI has specifically on if you use AI for health frequently, are you more likely or less likely to, you know, go to the doctor? And I still think it's too early to, um, you know, show the results of that.

Some will be thinking about their health, I think, much more [00:20:00] frequently, right? Because they're kinda conversing with this chatbot about what's going on with their health. That might give them heightened anxiety and think that they always need to go to the doctor, or that they need to talk to their doctor more often.

On the other hand, um, some might- To get the information from the AI and kind of be scared off to go to the doctor, and be like, "I don't wanna get bad news," or something. And then you have kind of another even bucket that's of people who, um, you know, think that they get the, uh, information that, uh, from the f- the information from the AI that, uh, maybe tells them, "Well, you know, I, I get the advice that I need.

I don't need a doctor's ad- advice on top of that." So I think there's gonna be people, uh, kind of spanning across those three categories, and it's still too early to say, uh, what the impact, uh, of AI health information will have on people actually going to the doctor. Mm-hmm. But overall, I'd say the doctor's role just, yeah, it's becoming a little bit less about the sole source of health information advice, and maybe more [00:21:00] about the trusted relationship, as Beth said, providing the human care and the clinical judgment that AI at this point can't reliably provide.

Yeah.

Marcus Johnson: Uh, all right, gang, let's, let's end by looking to the future. Really great presentation, uh, from April of last year, Frost & Sullivan, uh, Innoversa, um, presentation, looking at AI trends in healthcare. And this one slide they had looked at the, uh, different parts of the patient care journey. And so my question to you is, um, what area today do you think is being affected the most by AI, has the bus- biggest impact, and what area, uh, say, in a year or two's time down the road, will be impacted by it the most?

Uh, a couple of, just to give you an example of what the patient care, uh, journey looks like, uh, patient, patient triage. Um, it could be scheduling. Uh, AI can be involved in the clinical encounter, so is it, uh, helping to take notes, uh, through the data part of it, virtual assistant, maybe it's, uh, communicating with you, um, at home once you've left the office.

[00:22:00] Um, it could be used for, for imaging, diagnostic tests, surgical robots, revenue cycle, uh, the billing part of it, or maybe remote monitoring. Beth, what, uh, way do you think AI is impacting, uh, the patient care journey the most now and later?

Beth Snyder Bulik: I mean, I think probably on the, on the back end with the sort of the office, the administrative, um, role that AI plays now, so note-taking for doctors or, um, how, how doctors are using it on the back end maybe to do their own research.

I, I guess I think in the future maybe on the doctor side, um, using AI more often in diagnoses. They're relying on it more for those diagnostic or treatment decisions.

Marcus Johnson: Mm-hmm. Yeah. Rajiv, how about you?

Rajiv Leventhal: So for now, I think the most, uh, impactful way or what we're seeing, uh, as, as the greatest use is AI really, and I'll focus on the consumer side, AI really being like the front door of healthcare, right?

So traditionally- Mm ... [00:23:00] people, i- i- if something's, if something, you know, you don't feel good or you, you think you're, you have new symptoms and you're sick, you go to the doctor, right? That's just been like the traditional way of how life has worked. I think that AI is going to maybe if not replace, at least supplement some of those, um, kind of front door type services because you're going to ask an AI tool about your symptoms, you're gonna explain how you're feeling to the, to the AI tool, and whether, you know, it, it, uh, impacts whether you go to the doctor or not- That's to be determined, but I think becoming that really front door of the patient journey is, uh, where it's having the most influence n- now, now and in the near term.

Uh, in a couple years from now, I think we're gonna start seeing more advanced agentic capabilities that allow for things like direct appointment booking. We're just starting to see that.

Marcus Johnson: Mm.

Rajiv Leventhal: A recent partnership from Zocdoc, uh, and Gemini essentially allows, [00:24:00] uh, people to search for available care based on, like, their insurance details, what doctor- which doctors are in network, what their specific need is, if they have to see, like, a certain specialty.

And you can book that, and Zocdoc allows, like, the integration in Gemini. You can book directly in the Gemini portal, or the Gemini, uh, platform, without ever having to leave. And that's- Mm ... huge because Zocdoc is the leading healthcare booking marketplace out there. So I think a lot... we're gonna see a lot more of that in terms of just being able to do some more of the health-focused transactions directly in the AI tools, although we're probably a little bit of ways away from that.

But those agentic capabilities I think, um, are gonna be really fascinating to, to monitor.

Marcus Johnson: All right, gang, time for our top takeaways. Uh, I think one is, Rajiv, what you just said, AI is the new front door to healthcare. I think it's a really good one. Number two, why American patients are using AI for health reasons.

Uh, according to Zocdoc, 36% said to [00:25:00] prepare questions for a doctor. Close to 70% of patients have used AI after a doctor's visit, according to a Tembra survey. And then number three, people are now more likely to act on the advice from AI tools than they ever have been. Amongst AI users who asked a chatbot a health question, 81% acted as a result, according to a Rock Health survey.

Rajiv's full report is called AI Health Usage 2026. Uh, there's a link in the show notes, or you can head to emarketer.com if you're a Pro Plus subscriber. Uh, that's all we've got time for for today's episode. Thank you so much to my guests. Thank you to Rajiv.

Rajiv Leventhal: Thanks, Marcus.

Marcus Johnson: And to Beth.

Beth Snyder Bulik: Thanks, Marcus. Great to see

Marcus Johnson: you.

You too. Lance on the production crew, and thanks to everyone for listening in to Behind the Numbers: An EMARKETER Podcast, made possible by Atmosphere TV. Uh, subscribe, follow, rate and review, all that stuff. Suzy will be here Wednesday with the Reimagining Retail show.