It’s been said that 40% of all businesses will die in the next 10 years… if they don’t figure out how to change their entire company to accommodate new technologies. Large language models are driving the next generation of AI with predictions of massive industry disruption. But like any new technology, this new class of generative AI is raising issues of trust along with the need for more data types–especially in highly regulated industries like healthcare. Join Greg and Daniel as they host Dr. Andreea Bodnari from UnitedHealth Group to explore the pervasive use of generative AI across the healthcare ecosystem, and possible impacts across other industries.
It was a great conversation and one you don’t want to miss. Like what you’ve heard? Check out Episode One of The Main Scoop, Episode Two of the Main Scoop, Episode Three of The Main Scoop, Episode Four of The Main Scoop, Episode Five of The Main Scoop, Episode Six of The Main Scoop, Episode Seven of The Main Scoop, Episode Eight of The Main Scoop, Episode Nine of The Main Scoop, and Episode Ten of The Main Scoop, and be sure to subscribe to never miss an episode of The Main Scoop series.
Or stream the episode on your favorite platform:
Don’t Miss An Episode – Subscribe Below:
Disclaimer: The Main Scoop Webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we do not ask that you treat us as such.
Transcript:
Greg Lotko: Hey, folks. Welcome to The Main Scoop. Delighted to be back, and we’ve got an exciting topic for you today. We’re going to be talking about artificial intelligence and machine learning. We’re going to talk about the application of it today and get some perceptions and perspectives on where we think it is now and where we think it’s going to go. Generative AI and augmented AI. Does generative AI replace augmented AI? Is it instead of, is it in addition to? What do you think, Daniel?
Daniel Newman: Well, first of all, welcome back. I missed you.
Greg Lotko: I’m delighted to be back.
Daniel Newman: I think it’s important you came right back, you fired straight away, and we’re onto the show. Do you remember last November generative AI even being a topic? Did you talk about it much?
Greg Lotko: No, no, no. Nowhere near what you’re hearing today.
Daniel Newman: No. Didn’t talk about it at all. Not at all. I mean, now, frankly…
Greg Lotko: Aware of it.
Daniel Newman: We’ve been using it in different capacities, meaning the way you would interact with your Echo, and you were starting to see things like multi-turn conversational AI that was going on. Perhaps some of the filtering and recommender engines, the way they were interactively predicting was a version, early version. Or maybe you were using something like Google Workspace. And you know how when you start typing a sentence?
Greg Lotko: It’ll complete your thought?
Daniel Newman: It’ll complete your thought. And so, things were generatively happening.
Greg Lotko: Sometimes correctly. Sometimes as a help, and sometimes as, oh my God, that’s not at all where I was going.
Daniel Newman: Here’s the thing. Generative AI is going to find its way into everything we do. And it’s going to impact industries that are highly regulated, for instance, financial services.-
Greg Lotko: I’m glad you went there.
Daniel Newman: I mean we’re going to be talking to our bankers probably through some sort of text prompt and stream. And ideally, there’s going to be a very secure way for them to be able to see all your info and interact. And by the way, know they’re interacting with you. Those kinds of things are there. But we do have a big trust gap that we have to fill before we start moving from…
Right now, it’s open internet data. Meaning the generative AI that we’re experiencing is open internet data. That’s where we’re starting. But at some point, all the enterprise, the unique system of record data, the ambient data that exists inside of ecosystems that’s being collected in smart cities. That stuff’s going to get used, and it’s going to be used to generate a next level of experience. But we’ve got a trust gap, a massive regulatory gap, a policy gap, and of course, a attribution of content gap that I have no idea how our regulators that still don’t know what Google is.
Greg Lotko: Let’s talk about trust.
Daniel Newman: Two things. One is it’s going to be the unique data sets, not just the open data sets that’s going to make the difference.
Greg Lotko: Agree, agree.
Daniel Newman: And two, is a little bit like cloud. We’re going to find out that it’s not the panacea, but it’s going to change every industry. That’s absolutely the case.
Another industry it’s going to change big, and it’s going to be hopefully the focus of the rest of our show, is going to be healthcare, another highly regulated industry. Another one that you have to imagine, I’ve heard for a long time, are we going to have doctors doing our surgeries or is it going to be robots? Who will make our healthcare decisions? Are there ways to knock down the doors of HIPAA that have long made it hard to transmit and move information around about patients? These are big things that generative could help solve, but maybe rather than you and I talking about it, we should bring an expert.
Greg Lotko: Absolutely. I’m going to bring in the expert, because that’s not us. But I do want you to be thinking about something as we’re going through this conversation.
Daniel Newman: Well, about the tech, we’re the experts. But about the industry, I think we could do better. We could do better.
Greg Lotko: When we close out, I’m going to give you the final question now to close out so that you can ponder it now. I want to know if you believe generative AI is going to replace industry tech analysts. But we’ll get back to that.
Daniel Newman: I sure hope so.
Greg Lotko: Joining us today on The Main Scoop is Andreea Bodnari. And she’s a doctorate in AI and machine learning, and a senior executive at UHG. We’re going to move into the healthcare space, which inevitably, all of us need to worry about that at some point in our life. I recently did, with my broken leg and ankle.
Daniel Newman: Can’t even tell. That robot did a great job.
Greg Lotko: It does a fabulous job when I’m sitting. But Andreea, bring you into the conversation. I’m curious to hear your thoughts on AI machine learning in this space.
Daniel Newman: Everything we just said. What are we doing?
Andreea Bodnari: Well, first of all, thank you for having me. I’m so delighted to be here. And it’s very interesting to hear different perspectives on generative AI. Some of the applications we’ve seen so far of generative AI are more in the consumer space. And over there, there’s just a lot of perceptions that people have brought to bear. As I heard you talk, it got me thinking about a novel application in healthcare. When we typically go see the doctor, the doctor tends to give a diagnosis and a treatment. And a lot of that is protocol based. A lot of the diagnosis…
Daniel Newman: Patterns, right?
Andreea Bodnari: Exactly. What have I seen in the past? What does the medical society recommend? That’s association of data points, patterns, and then we take care of you. Hopefully, you get better. But there’s interesting instances where, for example, someone looks like they’re developing Alzheimer’s. And actually, they have a Lyme disease. And that type of unique instance, the type of unique association, is not something that a human would normally land on, a doctor will normally land on, in terms of let’s do a test to see if you have a Lyme infection. And let’s treat you for Lyme, because Alzheimer’s is a side effect.
But with some generative AI, this type of novel associations could be brought to bear. I feel like that’s a fascinating future. Of course, whenever you deploy AI, some form of automation on the clinical side of healthcare, you go through regulatory compliance. And that’s a long road.
We’ve seen some shortcuts being put forward by the FDA, just because they want us to make progress as a society. They see the inefficiencies and the typical way of regulating new technologies in healthcare that takes decades is not really going to get us to a paradigm shift. But it’s going to be the regulation that comes in the way, for sure.
Greg Lotko: I think that’s part of why you still see a lot of the AI or machine learning or the data that’s being mined, having to be certified or stamped at the end by a doctor or clinician that’s involved.
Andreea Bodnari: Correct.
Greg Lotko: Because there’s a desire in order for it to be certified and regulated, to have a human involved that has gone through the process. Do you see a future where we move past that? I mean, Chicago Med, I think, is a show that has OR 2.0. And they’ve got this AI that’s advising the doctor and telling them whether or not this can or can’t be done. And of course, it’s a TV show, so it’s like, “Oh, I’ve got a greater idea.” But is that where we’re headed, where the AI is going to be directing the human versus involved or assisting?
Andreea Bodnari: It’s possible. It feels right now, based on what we’re seeing in the industry, there is a couple of things that will have to fall into place. First and foremost, doctors do a lot of explaining back to the patient.
Greg Lotko: The good ones do.
Andreea Bodnari: Exactly. But coming up with a diagnosis is maybe seconds in a doctor’s mind. And then explaining the diagnosis, reassuring the patient, is the bulk of the job, the bulk of what the doctor does today. And if we’re able to improve on our explainability of AI methods, we could be at a point where that type of handover of the process to full automation happens. If-
Greg Lotko: You envision it as the AI doing a better job of the human communication and explanation of the condition than a doctor?
Daniel Newman: Or more thorough?
Andreea Bodnari: It’s hard to measure. I wouldn’t say it will do a better job, but there’s going to be instances where we can have comparable performance.
Greg Lotko: Okay. And are you on the dimension that he was going to, on being more thorough or more complete, not forgetting a portion of it?
Andreea Bodnari: I feel like that is an area where AI can do a better job. Humans get tired. AI won’t. Humans will have to optimize for 15 minutes or however long they have with you. And AI can stay there on the line forever.
Daniel Newman: But that’s what I wanted to get at though, is the general consensus of the consumer of healthcare does not find the experience to be particularly pleasurable. Now granted, if you have some sort of illness or disease…
Greg Lotko: With the doctor or with the AI?
Daniel Newman: That’s already problematic and that’s already… No, just in general, like patient experience.
Greg Lotko: The whole healthcare is…
Daniel Newman: The whole healthcare system is, first of all, how much actual interaction and time do you get with your doctors in general?
Greg Lotko: Not much.
Daniel Newman: They come in and like I said, you have a whole cadre of people that will be helping you, none of them of which are the ones that can actually fully diagnose you. Then you finally get the diagnosis. And it’s, “I got five minutes to spend with you to tell you something ho horrible is about to happen in your life. And then I have to, by the way, move on and do that again for another person.”
I think we also have this other thing that’s been going on over this is with self-diagnosis. The internet has become the WebMD. And good idea, but I’m saying you go down a rabbit hole. I got a weird mark on my hand. And you take a picture and you put it in Google and it’s like, “Oh, you’re dying of stage four liver cancer. That’s what that means.” You go, “Hold on.”
Greg Lotko: You’ve got days to get to the doctor.
Daniel Newman: Hold on, hold on, back up. There’s a lot of things that need to happen before this little mole that you think looks irregular means you have cancer.
Andreea Bodnari: Yeah. I want to go back to what you called out about the consumer experience in healthcare. Typically, when we think about the healthcare experience, we tend to think about the digital experience. You don’t think, “Oh, I had the bad experience,” because of how the doctor treated you. I feel like on average, doctors are mindful of how their communication lands on you. They want you to have a good experience.
Daniel Newman: It’s very limited. You get very little access.
Andreea Bodnari: It is limited. And most of your experience is going to be digital. Now, there’s been companies, startups as well as enterprise initiatives to fix the consumer experience in healthcare. But what we’ve seen is that the ROI for fixing consumer experience in healthcare is not there. Because in other industries that are consumer-driven, loyalty is a thing. If you have a better consumer experience, I’m going to come for example, shop with you than shop with Greg because I like the experience. I’m going to pay maybe a premium for that experience.
Greg Lotko: You’ll definitely pay a premium for it. Definitely.
Andreea Bodnari: Probably.
Daniel Newman: Rumor has it.
Andreea Bodnari: But in healthcare, you don’t have a choice more or less. You get your insurance through your employer. Your employer presents you with a number of options. If you move states, you are going to have to switch to another insurance plan. And again, you’re going to have a limited set of options.
And typically the consumer probably changes the insurance company year to year. And your choice factor is rarely the experience you have with this insurance company. It’s going to be, what’s my premium? Do they have doctors in the network? Do they have good coverage for the conditions I’m concerned? Maybe physical therapy is my main thing.
It’s a bit of an interesting dynamic that fixing the consumer experience in healthcare doesn’t translate to ROI and that’s why it hasn’t been fixed yet. There’s definitely ways to fix it, from a technology standpoint at least.
Greg Lotko: And there’s the access to the office or the information when you got to go through a call center. I find bots infuriating. The computer generated-
Andreea Bodnari: We got to give some credit to those bots. I mean, they pick up accents pretty well. They pick up my accent even when the dog is barking. I like kudos to that type of noise filtering.
Greg Lotko: Okay. They can understand you, but they still take forever for me to get to the answer to the question I want answered.
Andreea Bodnari: But right now, we might see a paradigm shift, especially in contact center operations because generative AI can humanize more of that experience. I feel like getting the right answer, part of it could be just better information architecture.
Greg Lotko: I agree.
Andreea Bodnari: But once you know what’s the information that has to be delivered, right now it’s a dull delivery. And with generative AI, you can package it a bit better. And you can think more about what’s the perception factor? What’s the sentiment? Is this consumer coming in frustrated and can I make them happier? Do they want to refill? And because they’re running out and they didn’t think they’re going to run out so fast. And can I reassure them that even if they skip a couple of days, it’s fine because I can read that in the medication prescript.
Greg Lotko: Yeah. I see opportunity. I know there is because it’s about the data. The data set of not only knowledge of what’s going on, but of what the human interaction is. And when somebody uses these keywords, what they’re more likely working for. I believe my issue with it is more a lack of knowledge in the data set. And it seems like the bot never has the path for the question I’m asking.
Andreea Bodnari: The good news is that interoperability is way more practical in healthcare than it used to be even two years ago. And we have both enterprise initiatives that have pushed interoperability to a point of being pragmatic. We also have a lot of startups that have opened up the ecosystem for new digital applications. You talk a lot about having access to data that sometimes is siloed.
Having access to more data doesn’t necessarily mean that you’re better off. Some of the data you would get access to is very noisy. Which means that you cannot really take it as a source of truth. It’s very humbling to actually see that clinical data in general tends to be noisier than, for example, claims data or pharmacy data.
And clinical data is what you would want to make personalized recommendations or to speed up discovery of new treatments. At the end of the day, there’s no panacea. You have the technology advances right now that make interoperability possible in healthcare, but you still have to think focus. What’s the sandbox where I’m going to stitch things together and start to make information flow possible?
Greg Lotko: And it’s driven by the challenge, the problem or the opportunity that you’re trying to address.
Andreea Bodnari: Absolutely. Absolutely.
Greg Lotko: And there is a lot of noisy data out there is a lot of billing and claims data that has gone through the mainframe platform that tends to be more well cleansed, more reliable, but it is about the interaction across multiple sources. It’s not just this silo. It’s about federation, data mining, protection, security, and then the insights that you can drive from it. For sure.
Daniel Newman: I’d love to kick this back to Andreea to take us home here a little bit.
Greg Lotko: Sure.
Daniel Newman: Obviously, you have a lot of general perspectives. You’ve studied the AI industry all the way through to your PhD. You’re watching, you’re observing. I’m sure you’re thinking and you’re being asked to consider the best ways to implement. What’s your overall take? When does this really take hold? What’s the future? Any thoughts about how this is going to evolve?
Andreea Bodnari: Lots of good thoughts. I’m going to stick to the highlights over here. But since we talked about the M word over here, I feel like there’s a very strong application for mainframes in healthcare, specifically in hospital settings.
If you have to think about the surgical hospital or even a hospital that has a trauma center. If there’s an earthquake and you can no longer connect to something like a cloud, then you still need to have access to the patient record to see what medications they’re on or what type of conditions they have. You would want that data to be locally stored on a mainframe, well encrypted. I feel like mainframes are definitely well suited for us to manage through disasters and to have a fallback in certain high critical settings.
Now, going back to AI and how long it’s going to take for it to be implemented in healthcare, we are seeing good progress already on the administrative side. When you start to solve for business workflows by either augmenting or automating them on the administrative side in claims processing and managing prior authorization submissions, there’s point solutions that are backed by AI that are being adopted at scale.
Now, in healthcare in general, technology gets adopted through regulation. It’s not like someone comes up with an innovation and everyone is on it. It’s the best thing. It just gets deployed at scale. It tends to be a slow moving train just because the industry is risk averse.
We are working and hoping that the government will start to put more regulatory frameworks in place that give healthcare stakeholders some reassurance about where they can deploy AI and some guidance about how to deploy AI. We can transition from the administrative into the clinical side.
I feel like on the clinical side is where there’s the most need and the most potential for AI. But at least we’re building a bit of status quo by solving for administrative applications. And by the time the regulatory framework enables deployments in clinical, we’ll be ready to sail at full speed.
Daniel Newman: And she got a sailing reference in, Greg. Andreea, thank you so much for joining us here on the Main Scoop today.
Andreea Bodnari: Thank you for having me. It was my pleasure.
Greg Lotko: It was a pleasure having you. I hope you folks enjoyed this. It really was a topic about artificial intelligence, but having Andreea here, we had our own AI. She’s actually intelligent.
Daniel Newman: And is she a bot?
Greg Lotko: She’s not a bot.
Daniel Newman: She’s not a bot.
Greg Lotko: She’s a real person.
Daniel Newman: Hey, so you told me at the end I’m going to have to answer that question. I want to take you up on that.
Greg Lotko: Well…
Daniel Newman: I’m building a GPT. Yes, I actually genuinely believe that most information sources…
Greg Lotko: Will you continue to get paid?
Daniel Newman: I don’t know.
Greg Lotko: Or will the GPT be paid?
Daniel Newman: Well, I actually, I’m rerouting. I’m rerouting all of the payments using a generative to a new routing number that my bot has set up. And I’ll just suddenly be broke and I won’t know what happened. There’s a kill switch operator job opening up. When this is over, and we have to hit the, oh my god, hit that kill switch so we don’t end up in 2001 in Space Odyssey. And, “No, I won’t open the door.”
Yes, you will open the door. And I know we could go on and on with this. Let’s just put it this way. I think there was a data point that came out that I think 40% of the Fortune 1000 CEOs think by 2030 that their business models will be completely disrupted. They will no longer be relevant. That’s within five to six years. I can’t do math. I’m doing math.
My point though is that it’s something that came from pretty much nowhere in six months has made most of the Fortune 1000 CEOs say, “Our business model will no longer be relevant.” Greg, that’s crazy. That’s awesome. That’s why we brought this topic to bear, Greg, is I mean mainframe at the core will be a big part of the generative AI story.
Greg Lotko: I agree.
Daniel Newman: But in the end, this is going to change every business. It’s going to change every industry. And hopefully all the people that are part of our Main Scoop community have enjoyed this show. Hey, everybody out there, hit that subscribe button. We loved having you as part of our main scoop. Greg Lotko, welcome back. We missed you.
Greg Lotko: Happy to be here.
Daniel Newman: For Dan, for Greg, for myself, for the Main Scoop, we’ll see you all later.
Author Information
Daniel is the CEO of The Futurum Group. Living his life at the intersection of people and technology, Daniel works with the world’s largest technology brands exploring Digital Transformation and how it is influencing the enterprise.
From the leading edge of AI to global technology policy, Daniel makes the connections between business, people and tech that are required for companies to benefit most from their technology investments. Daniel is a top 5 globally ranked industry analyst and his ideas are regularly cited or shared in television appearances by CNBC, Bloomberg, Wall Street Journal and hundreds of other sites around the world.
A 7x Best-Selling Author including his most recent book “Human/Machine.” Daniel is also a Forbes and MarketWatch (Dow Jones) contributor.
An MBA and Former Graduate Adjunct Faculty, Daniel is an Austin Texas transplant after 40 years in Chicago. His speaking takes him around the world each year as he shares his vision of the role technology will play in our future.
