AI Adoption and Skepticism in Regulated Industries - with Ylan Kazi of Blue Cross Blue Shield of North Dakota
The AI in Business Podcast · 27 Jan 2026
Source: https://podcast.emerj.com/ai-adoption-and-skepticism-in-regulated-industries-with-ylan-kazi-of-blue-cross-blue-shield-of-north-dakota
Transcript: https://podcasts.opennash.com/shows/ai-in-business/ai-adoption-and-skepticism-in-regulated-industries-with-ylan-kazi-of-blue-cross-0c7c05/ (via OpenNash Podcasts, machine-generated)

Welcome everyone to the AI in Business podcast. I'm Matthew Demello, Editorial Director here at Emerj AI Research. Today's guest is Ylan Kazi, Chief Data and AI Officer for Blue Cross Blue Shield of North Dakota. Ylan joins us on today's show to discuss balancing AI innovation with risk in regulated healthcare, framing risks realistically compared to medical errors and human inaccuracies. Our conversation also covers practical workflow changes, including cross-functional governance teams for policy development, starting AI experimentation with low-risk use cases to build standardized processes, and leveraging AI to improve patient experience, such as lab result explanations and wait time predictions.

But first, a quick message from our sponsors. AI agents accelerate workflows, but errors can multiply fast. Rubric Agent Cloud provides full visibility, enforces policies, and rewinds actions in minutes. It runs continuously, giving guardrails, tracking activity, and providing a safety net so teams can scale AI without risking critical operations. This segment is sponsored by Rubric Agent Cloud. If your business relies on AI agents, you can get exclusive early access to monitor, govern, and rewind their actions at Rubric dot com.

That's R-U-B-R-I-K dot com Rubric dot com. Interested in putting your AI product in front of household names in the Fortune 500? Connect directly with enterprise leaders at market-leading companies. Emerj can position your brand where enterprise decision makers turn for insight, research, and guidance. Visit Emerj dot com slash sponsor for more information. Again, that's E-M-E-R-J dot com slash S-P-O-N-S-O-R. Without further ado, here's our conversation with Ylan.

Ylan, thanks so much for returning to the show. It's phenomenal to have you back.

Yeah, thanks, Nick. Great to be here again.

Yeah, so Ylan, in highly regulated sectors like healthcare, the conversation around AI isn't just about innovation, it's about risk too. Whether that's the inherent risk in data handling and exposing medical records or PII, or bias in trials to model regu- uh reliability, regulations, et cetera. There's a lot of risk there, but there's a paradox too. Um, many leaders face this paradox that extreme caution can actually become its own form of risk. So while compliance and safety are currently dominating the dialogue, it's kind of easy to overlook the fact that if you delay adoption, it could cost you even more in lost opportunity, inefficiency, in patient outcomes, um, than taking those calculated and well-governed steps forward.

So, with all of that being said, what is your take on risk aversion, and how do you see it shaping AI adoption in healthcare and other regulated health sectors right now?

That's a great, great question, and I think you're right. When we look at healthcare in general, in the different sectors, we're working with some of the most sensitive data in the world. Patient data, member data, health conditions. And so there there always is just that that very high bar and that high standard for making sure that we're keeping PHI and and and PII very, very safe and secure. When it comes to AI adoption and some of the the safety conversations, the regulation, there are elements where it is always going to differ from other industries.

You know, if we look at at retail, for instance, one of the worst things that could happen is somebody gets ad targeted incorrectly and, you know, they don't wanna buy that T-shirt. With healthcare, right, obviously very different. I think that what we're seeing and and what I've been seeing very recently is, you know, we we do need to still be talking about safety and governance, but in a way that enables innovation and adoption. And I don't think that those two things are necessarily mutually exclusive.

I think in our minds, and especially for those that have been in healthcare a while, sometimes we can have, like, self-limiting beliefs. But I think this is really an opportunity to to push the envelope. And when it comes to risk aversion, we we can't be at the point where we just don't apply AI to healthcare. That's way too risk-averse. I think what we need to be doing more is actually detailing out, okay, what are those risks that we have in our minds?

And then actually looking at the data or looking at the use cases to really see, does that make sense, you know, in in a a realistic, um, standpoint. And one of the examples that I use a lot is when we look into healthcare, you know, medical errors and and harm and death from medical errors, it's pretty high. It's it's in the hundreds of thousands of use cases, you know, potentially even more. And yet, we still practice medicine, we still do surgery, we still do all these functions with an understanding of the risk and putting safeguards into place.

And I think that's how we can also treat AI so that we get all the benefits from it, while also still, you know, reducing any potential risk or even reducing any potential harm.

Absolutely, absolutely. Now, from your perspective, what do you think is kind of a better way for leaders to kind of frame those risks and the benefits of AI adoption? You know, there are so many ways of applying this technology and and approaching the risks. But from your perspective, having been in the industry for such a long time, um, what are you seeing? Where are you seeing victories in in per perhaps colleagues, perhaps, um, clients? Where are you seeing better ways for leaders to frame the risks and and the benefits?

I think first off, just by talking about it in depth and using examples and using analogies. And so one of my one of my absolute favorite questions to, you know, ask those individuals that are maybe skeptical, or, you know, that are very concerned about AI risk is, you know, asking them, what do they do on a daily basis? That is probably the most risky behavior. And for many of us, that's driving, right? Driving is inherently more dangerous, more risky than most of the things that we do on a daily or weekly basis.

What we don't do, though, is we, you know, we don't get into our car, we don't put ourselves in a reinforced bubble and go twenty miles an hour so so that we won't get in an accident. But we do take some level of caution and some level of risk reduction. We we try to drive the speed limit, we wear our seatbelts, you know, we use lane indicators, we use on-ramps, off-ramps. So I think that's a good analogy towards how do we treat AI and the adoption of AI.

You absolutely have to have the right governance standards, guardrails in place, but you don't wanna constrain it so much that you don't see the value. And, you know, most importantly, the benefits to members and patients. And I think that's a that's a very helpful way to think about it because there's also the element that anything that's perceived as AI automatically will be treated with a much higher standard than a corresponding, you know, human that is doing the same role.

And what I find is that we're a lot more accepting of humans making errors than AI. And, you know, like, if if an AI has 95% accuracy, you can bet that whoever you're talking with, they're gonna say, well, it's it's gonna be wrong 5% of the time, you know, should we really think about it? And my counter to that is, what is the human equivalent, right? So people, many times, are incorrect. Some people can just blatantly lie, and yet we still all work together and we still kinda have that good understanding of that is the human condition, and we can strive for excellence, but there is always gonna be some level of error.

And I think we can apply that also that mindset to how we're using AI and and how we really, how we really adopt it. The other one that I'll I'll point out is we're using AI for, you know, very, very complex industries. So the the other example I think about is, you know, Waymo and and the self-driving cars. Waymo is outperforming most human drivers, you know, over a over a set amount of miles. So it's already showing that, you know, actually even driving, which is inherently dangerous, the robots are doing a lot better on average than your average human driver.

And so I think as we start to see more of these studies, as we as we start to have more of a longitudinal data, that's also changing the the culture of how we think about AI and how we use it, you know, in our our daily lives.

Absolutely. Um, yeah, that automotive, uh, Waymo and your automotive metaphor, um, seatbelts, et cetera. I was wondering, you know, from your perspective, you mentioned, you know, we're driving around in our reinforced bubble, um, we put on our seatbelts, we're cautious, et cetera. Um, but we also have a lot of road signs about, we have traffic lights, we have traffic officers. Um, just wondering, uh, who needs to come to the party here? Um, you know, we've got the the the healthcare side, the the providers, um, the researchers, et cetera.

Um, and then we have the regulators. And at the moment, with the influx of AI, um, it doesn't quite seem that we have all of the road signs in place, the the traffic officers in place. And I'm just wondering how, um, how leaders should kind of consider the, uh, regulatory side of things and and the fact that things are still pretty gray at this point in time.

Yeah, from a from a regulatory side, you know, the the challenge that we're seeing at a both a federal and a state level is that generally any anything around AI safety, AI governance, it's all, was very well meaning. So I think that's that's how it begins, right? Like, really thinking about here's the art of the possible, you know, in a perfect world, what are all the things we would want in place to drastically reduce risk? But I think in many cases, that runs into challenges on how businesses actually perform in the market and how they, you know, how they want to use AI to bring more value to their customers.

So I I think what I've seen is being able to have a good cross-functional team is extremely important. You know, you want to have legal at the table, you wanna have finance, HR, governance, et cetera at the table, so you can have these robust discussions before you're creating many of the policies and procedures. Because otherwise, what you see is you see these these perfect academic policies that really have little or no applicability in the real world.

And that makes it very challenging. I think the other area is having having more of those innovators and builders at the table, as well as the frontline staff that are gonna be utilizing the solutions. I think in many cases, you know, their voices aren't adequately heard. And so I think what that does is it overemphasizes, you know, the the regulators or or the AI safety experts. You know, you hear from them a majority of the time, but the problem is that many of those individuals have never actually implemented AI.

They don't know how to go about doing that, and so they have limited knowledge in that in that aspect. And I think once again, that kind of tilts a little bit too far on the safety side. Like, you have to have that cross-functional group at the table. If you are going to balance those competing needs and do what is best for your organization and for the customers that you serve.

Right, right. Yeah, I guess the more people at the table, they say, too many cooks spoil the broth, but I think in this instance, the opposite is the case. But changing tack a little bit, um, you've spoken plenty about, even on this very show, about the Hippocratic oath, um, but to perhaps maybe tread a slightly different angle on the topic, how do you think it should shape experimentation and governance in healthcare AI?

Yeah, I'm I'm a big proponent of what I call the AI Hippocratic oath, and I've, uh, it's been no secret, I mean, on on this podcast, on others, um, very vocal. And I think it helps, especially in healthcare, to think about it that way because we have, you know, your standard Hippocratic oath of do no harm. And it's a great analogy because, you know, any time, any time a physician is doing a surgery or performing a procedure, there is always that potential for harm.

But what they're not gonna do is sit there and not do anything because that would, in actuality, be a lot more harmful. And so I think that's a good, it's a good way to think about how that shapes the experimentation, how that shapes, you know, different types of innovations. Because you don't have to apply AI to those moonshot ideas. In fact, what I would really recommend is starting with those easier use cases or the ones that are more manageable.

Because a lot of AI, it's as you're applying it, as you're adopting it, you're going to make mistakes along the way. But if you're doing that in a safe setting and, you know, in small enough use cases, you can iron out a lot of those, you know, issues and challenges that come up. So that when you do work on those use cases that are broader in nature or that may present a higher level of risk, you have more of a standardized process, um, to go about it.

I think the other piece is also quantifying what is what is that actual risk. So, for instance, if I if I walk outside, my risk of being struck by lightning is extremely small. Like that is a very, very small tail risk. There are corresponding be brought up, you know, not the lightning strike, but something equivalent that, you know, will happen one out of a million times. And it's like, should you actually shut down the use case because of that one out of a million, or should you appropriately quantify what could be what could be that downside?

And I think there's a big difference between downside risk versus more of, like, that catastrophic risk that you just would never want to, you know, go about applying AI to, or where where one makes sense to do.

Absolutely, absolutely, yeah, it's one gets the sense that this Hippocratic oath should be stolen from the medical field and perhaps applied across all, uh, AI vendors and and the big developers these days, uh, just to keep in mind the safety front and foremost, cause it feels like we could be on the edge of a slippery slope if things don't go right. So to continue the conversation, though, um, I guess I wanted to ask you from your perspective, how do I guess the cultural comfort, the the fact that we've gotten used to AI, it's becoming more prevalent everywhere.

How does cultural that cultural comfort comfort and the broader public perception, um, influence adoption in the medical and and, um, in the medical space, um, what are what are you seeing out there that is having a kind of to-and-fro effect between the public perception and actual applications in the medical field?

The the cultural piece, and I think just the comfort, familiarity with AI is going to drive a lot more of that bottoms-up innovation and capabilities. And what has been very interesting to see, especially with the advent of, you know, ChatGPT and other LLMs within healthcare, you know, many people now when they go to the doctor, if they receive their lab results or clinical notes, they literally take them, they copy and paste them into ChatGPT, and they have it explain to them, like, what is actually going on, you know, e- explain it in a much simpler fashion.

And that's just something that has been done very organically, but I think that patients and and members are finding ways to utilize this to help improve their overall patient experience. And so I, you know, I think that's it's it's an opportunity, you know, especially within the the provider side, to really lean into that versus maybe dissuading patients from from doing that. And, you know, I would obviously always caution anybody as you're using that, you know, don't take that as as gospel and and work with your, you know, your doctor, your nurse around it.

But I think patients are becoming more demanding around it because they've had amazing experiences, you know, from Amazon and other retail companies. They want that in healthcare, and so customer experience has been very high up on the list. I think the other, you know, the other thing that that patients and and the healthcare ecosystem runs into is around affordability. I think that AI could really be applied more intensely to help with some of the affordability challenges.

And even just, you know, some of the abrasion that that comes up, right? Like, I s I still go into my doctor's office and I have no idea if I'm gonna wait for five minutes and thirty minutes. That shouldn't be an issue anymore with AI. So I think I think there's going to be, you know, more disintermediation within the healthcare field, particularly from from AI.

Absolutely, yeah. I mean, I've I've recently heard someone talking about, or speaking about, the future prospects and what is the position of the doctor as we as as things progress. And um, with all of the medical knowledge available, um, at the tap of a button, it does put the the medical practitioners themselves in quite an interesting place, I guess. Now that, you know, they could kind of almost be seen as as purely as, I guess, custodians of this information.

And and if you look at a doctor as a custodian of information, well, how are they gonna hold onto the information better than a chat GPT could, for example? So I see a lot of interesting, at least from the practitioner side of things, um, a lot of interesting changes probably on the horizon, and I'm I'm super intrigued to see how they pan out. Ylan, it's been a pleasure having this conversation with you, getting a sense of on the ground, how the healthcare space is, um, approaching AI, some of those adoption risks, and how to, I guess, navigate the sort of skepticism that does come with a new technology.

Um, so thank you very much. It's been great having you on the show, and I look forward to having you back again.

Yeah, thank you, Nick. Great to be here again.

Wrapping up today's episode, I think there were at least three critical takeaways for healthcare CIOs and compliance leaders from our conversation today with Ylan Kazi, chief data and AI officer at Blue Cross Blue Shield of North Dakota. First, frame AI risks realistically by comparing to everyday human errors like driving, using safeguards such as governance without over-constraining innovation. Second, form cross-functional teams, including legal, frontline staff, and builders, to develop practical policies that balance safety and adoption. Finally, start experimentation with low-risk use cases to refine processes before scaling to high-impact areas like patient experience.

Are you driving AI transformation at your organization, or maybe you're guiding critical decisions on AI investments, strategy, or deployment? If so, the AI in Business podcast wants to hear from you. Each year, Emerj AI Research features hundreds of executive thought leaders, everyone from the CIO of Goldman Sachs to the head of AI at Raytheon and AI pioneers like Yoshua Bengio. With nearly a million annual listeners, AI in Business is the go-to destination for enterprise leaders navigating real-world AI adoption.

You don't need to be an engineer or a technical expert to be on the show. If you're involved in AI implementation, decision-making, or strategy within your company, this is your opportunity to share your insights with a global audience of your peers. If you believe you can help other leaders move the needle on AI ROI, visit Emerj dot com and fill out our thought leader submission form. That's Emerj dot com and click on Be an Expert. You can also click the link in the description of today's show on your preferred podcast platform.

That's Emerj dot com slash expert one. Again, that's E-M-E-R-J dot com slash expert one. We look forward to featuring your story. If you enjoyed or benefited from the insights of today's episode, consider leaving us a review on Apple Podcasts and let us know what you learned, found helpful, or just like most about the show. Also, don't forget to follow us on X, formerly known as Twitter, at Emerj, and that's spelled again, E-M-E-R-J, as well as our LinkedIn page.

I'm your host, at least for today, Matthew Demello, Editorial Director here at Emerj AI Research, on behalf of Daniel Fagella, our CEO and Head of Research, as well as the rest of the team here at Emerj. Thanks so much for joining us today, and we'll catch you next time on the AI in Business podcast.