Healthcare organizations don’t just have an efficiency problem—they have a culture problem. Siloed specialists, misaligned incentives, and fragmented decision-making leave patients frustrated and clinicians burned out.

Jonathan Kolstad is a professor of economic analysis and policy at UC Berkeley Haas and is one of the country’s leading health economists. He’s the founder and faculty director of the Center for Healthcare Marketplace Innovation (CHMI), a joint center between Haas and UC Berkeley’s College of Computing, Data Science, and Society. CHMI’s executive director is Ted Robertson, who specializes in designing and building healthcare products with the best mix of human and AI insights in decision making.

On this episode of The Culture Kit with Jenny & Sameer, Jon and Ted join organizational culture expert and co-host Jenny Chatman, dean of the Haas School, to explain why healthcare’s broken structure is ultimately a culture problem, and how AI—deployed in the right way—might help fix it.

(Note: Co-host Sameer Srivastava was out of town for this episode.)

3 Main Takeaways:

  1. Healthcare’s fragmentation is baked into the incentive structure, creating professional subcultures that work against patients and each other.
  2. AI has the potential to reduce burnout in healthcare providers and give patients a higher quality standard of care.
  3. General-purpose AI isn’t enough: healthcare needs models trained on real clinical decision-making, not medical textbooks.

*The Culture Kit with Jenny & Sameer is a production of Haas School of Business and is produced by University FM.*

Show Links:

Transcript

[00:00:00] Jennifer Chatman: We talk a lot on this show about workplace culture, but healthcare has a culture, too, and patients feel it the moment they walk through their healthcare provider’s door. Today, we have two guests, experts who are doing some of the most ambitious, cutting-edge work on the future of healthcare systems, and they’re doing it right here at Haas with collaborators across UC Berkeley.

I can’t wait to hear about what they’re working on, but first, let me do something I usually don’t do on my own and welcome you to the show. I’m Jennifer Chatman, a professor and dean at the Haas School, and this is The Culture Kit, where we give you the tools to build a healthy and effective workplace culture.

My co-host, Sameer Srivastava, is away today, so I’m flying solo, and I’m lucky to be joined by two extraordinary guests. Jon Kolstad is a professor of economic analysis and policy at Haas, where he holds the Henry J. Kaiser Chair. He’s one of the country’s leading health economists, and he’s the founder and faculty director of the Center for Healthcare Marketplace Innovation, a joint center between Haas and UC Berkeley’s College of Computing, Data Science, and Society.

Jon’s also an active entrepreneur and founder in healthcare and technology. Jon, welcome.

[00:01:21] Jon Kolstad: Thanks, Jenny. Really excited to be here and looking forward to talking about many pieces of healthcare, but particularly the culture side, which I think is so important.

[00:01:30] Jennifer Chatman: Joining Jon is Ted Robertson, the executive director of that same center, CHMI. Ted specializes in designing and building healthcare products with the best mix of human and AI insights and decision-making. He helped found the Coalition for Healthcare AI to help the industry develop guidelines for effective use of AI in healthcare.

Ted, it’s great to have you here.

[00:01:55] Ted Robertson: Thanks, Jenny. It’s a pleasure to be here.

[00:01:57] Jennifer Chatman: Okay, so I want to start with something that maybe many people have experienced. It’s going to sound like regular complaining about healthcare, but it’s actually, I think, an issue that will tee up some of the problems that you’re working on and trying to solve in the healthcare industry in the U.S. So, suffice it to say that I get blood tests fairly regularly, and I go to the blood test center. I’ve done all of the pre-work, filling out all the forms, all the insurance. I have an appointment, all the things, and I get there for my appointment time, and often I have to wait a half an hour, an hour, and I’ve waited so long that I’ve actually just gotten up and left without getting the test.

What is the deal? Why is what seems to be a straightforward, routine requirement for a patient so difficult to complete?

[00:02:57] Jon Kolstad: Yeah, I mean, I think, sadly, probably all of us have had some flavor of that with friends, family members, or ourselves. And I think, Jenny, you’re hitting on one piece, which is something that’s happened in healthcare for a long time, which is there’s not been the level of investment in being consumer-centric and, sort of, the kinds of operational efficiencies that we might see in other industries.

Though I think that’s not actually the main thing. I mean, the scale of healthcare is so big. The investment in operational improvement is real, and there’s been operations experts focused on throughput and the kinds of processes and software for a long time. Some of them are a little more legacy, but I actually think it gets to something a little bit deeper, which is something probably we can talk a lot about today. That, if you think about who the healthcare system has to serve, they’re serving everybody. All kinds of folks are getting care, and that care depends on a whole lot of different eventualities they might have, right? There’s lots of conditions coming in.

So, even if they know precisely what you have, there’s a lot of folks coming in with different insurance and different eventualities and different requests from their doctor, so there’s almost a level of clinical intelligence, right, if you had a doctor there or a nurse there reviewing everyone as they came in the door, you might end up with a pretty different flow, but that’s a big difference from just, let’s get the queue right and some software in place.

And I mean, we can tell you, I think this is true of labs. It’s true of, you know, both for patients and doctors, right? Doctors don’t like to be behind, and they’re given a full schedule that’s largely independent of how complicated those patients are going to be. So, this is a common phenomenon, sadly.

[00:04:32] Jennifer Chatman: Well, so let’s start with some context. I mean, healthcare organizations are structured very differently from most workplaces. You have physicians and nurses and administrators and payers, all of whom are trained with different professional approaches, and each of which operates under different incentives.

And as I understand it, some of the key specialists, surgeons, anesthesiologists are essentially independent contractors coming in for specific work and then leaving. So, can you walk us through that structure and how it shapes culture, or perhaps makes culture harder to build in the first place?

[00:05:11] Jon Kolstad: I think this is something that’s really interesting, and I think often, kind of, underappreciated, particularly from people looking at the outside at healthcare, that, you know, you look at a hospital, it’s a building, it has a name, you know, a company’s running it, it looks like an organization, and it is an organization.

But practically speaking, most of these healthcare providers have the operating rooms, they have the infrastructure, and then, in many cases, they’ll have a surgical suite, say, and the surgeon is being paid separately to do surgery in that suite.

And so, if you think about, you know, Medicare, which is the largest single payer in the U.S., they have a physician fee, and they have a hospital fee. In fact, explicitly, the healthcare system can’t pay the doctor directly because of regulations about anti-kickback and not wanting to encourage them to come into that space. And similarly, the doctor, in this case, can’t own part of the overarching system because you don’t want misaligned incentives about doing, say, more surgery.

So, there’s a bunch of features in the background that create a pretty fragmented organizational structure and also a situation in which you, kind of, have an internal marketplace, almost, where the hospitals, in this case, are trying to attract the surgeons, but with non-monetary things like surgical operating times, and the surgeons are trying to find where they can do the most work.

Now, there are integrated systems, Kaiser Permanente being a famous version of this, Geisinger, and they try to build their own cultures, but they’re still in this broader ecosystem of clinicians, sort of, really operating as kind of independent contractors. That’s very much in the ethos of the industry.

[00:06:48] Jennifer Chatman: So, Ted, what’s your perspective on this?

[00:06:51] Ted Robertson: So, I actually think that this question about how healthcare is structured and the fragmentation of it has very significant cultural ramifications, and it speaks to your opening in terms of your experience as a patient, how disjointed and fragmented you felt.

So, I think there is, one of the biggest problems that occurs is coordination breakdown. So, a lot of healthcare is about sequencing decisions over time across people, and when those people aren’t tightly aligned, or they have slightly different incentives, or they don’t have clinical knowledge and intelligence across that, you get gaps, you get things that are late, you get missed follow-ups, you get redundant or missing tests, you get people who ask you the same questions every appointment, even though they already know your story, and they still don’t know what’s going on with you.

So, there’s this coordination problem. I think number two is you get inconsistent decision-making, so you get two patients with very similar conditions who can have very different experiences depending on which path they happen to take through the system, and that’s not always about medical uncertainty. It’s just about organizational culture and inconsistency.

I think there’s another problem, a third problem, which is ambiguity around responsibility, that all this fragmentation creates these gaps, again, that you don’t really know who’s finally making that decision or how to push it forward. We’ve all probably had that experience where we’ve had to call or ask or say, “Is that really the thing I need to do, or is that… Or are you the person that can help me?” And it’s because there’s this ambiguity that comes from this structure.

And then the last thing I’ll say is, ultimately, not only is it not good for the patients, it’s not great for the clinicians, right? They have this huge cognitive overload where they’re constantly trying to stitch together a story and understand what’s happening, and they just have to deal with more and more data and backlog. How do we expect them to actually make good decisions with us in seven minutes? It’s a near impossibility.

So, it’s a big problem across the board, how this structure affects the culture.

[00:09:00] Jennifer Chatman: Yeah, it’s a serious list of challenges that you just ticked off there. So, Jon, as an economist, you study how incentives shape behavior in healthcare. Do incentives actually push doctors and nurses and administrators toward a coherent organizational culture, or do they actually pull them away from something unified?

[00:09:22] Jon Kolstad: I’m an economist, I have to say incentives matter, but I think in healthcare they’re very related to the organizational structures we see, but also, frankly, some of the way that patients move through the system, the way information flows, the way different peer clinicians might interact is very related to differences in the incentives.

So, you have often, kind of, a primary care clinician upfront, either they are a nurse or a doctor, they’re trying to coordinate, you know, understand what’s going on with the patient, manage care over time, coordinate where they’re going, and they’re referring out appropriately, given the segment and just the sheer volume of knowledge, right? You want specialists, you want specialization, you need to think about the healthcare system as a system-level intelligence, not just, you know, one Dr. House here solving everything that’s hard.

And so, patients are moved out to those specialists. Those specialists are paid on a largely different fee schedule. They’re paid a lot more per procedure or per activity.

So, once they’re sent to a particular direction, right, you send them to someone with a hammer, they’re going to be looking for a nail, right? You send them to a surgeon, they’re much more likely to get surgery, even if at the margin. Maybe if they’d been sent to someone who’s able to medically manage their pain, for example, they would’ve been more likely to get a medical management.

The reason it’s different, and the reason there are, sort of, in these incentives in each of these silos, is related to the payments, but there’s not an overall incentive to say, let’s manage this patient’s health holistically. I mean, I think in the background, there are important other non-pecuniary cultural institutions, Hippocratic Oath being, you know, the elephant in the room, but those are traded off against the realities of the incentive structure that our system has, which is to do more of what you’re paid to do within your silo.

[00:11:15] Jennifer Chatman: Yeah, I mean, that’s interesting. So, you’re an economist, I’m a social psychologist, and I agree with you that the incentives are really driving behavior in most situations. In this case, because the incentives are so distinct, what it means is that there’s no coherent overarching culture. There are really subcultures that probably follow the professional categories, which are based on the incentives that those different categories receive.

[00:11:46] Jon Kolstad: Totally. There’s actually an interesting aside on this. Our colleague here at Haas, Dave Chan, has a great paper. So, the way that those different payments that are adjusted every year are done by a committee, and that committee has representation from each of the different specialties, but it’s not exactly a majority vote. But you can imagine that surgical specialties, relative to the total volume, are heavily overrepresented, and primary care is underrepresented.

So, you, sort of, not only do you have these silos, but as these things are updated, there’s tended to be this drift towards rewarding procedural activities. Those are even more… I mean, they’re amazing for patients that need them, but they’re much more like, “I’ve got a hammer, let’s find the nail and then move you back into the system,” and much less holistic.

[00:12:35] Jennifer Chatman: Yeah. Well, and again, in the psychological world, we would talk about that as a status story and power shifts occurring. So, maybe we start to think about some of the brighter future issues that you’re working on, and I think here you can’t have a conversation today about healthcare without recognizing the immense potential of AI to really transform healthcare, perhaps one of the sectors that could benefit the most from greater integration of these powerful tools.

So, there is a lot of AI washing happening in healthcare. Vendors are promising transformation, hospitals are buying in. How do you separate the genuine breakthroughs from the hype, and if so, what does a real AI deployment look like versus a superficial one? Can you give us, you know, a concrete example of something that’s actually worked so far?

[00:13:35] Jon Kolstad: So, yeah, Jenny, I think it’s a great question. I think we’ve seen some really exciting applications and potential for applications of AI in healthcare. It’s certainly going to change a lot of the things that we’re talking about over time.

But I think healthcare is a really good example where, kind of, conventional wisdom and trying to bring general-purpose technologies into this, you know, 20% of GDP, that’s a very complicated part of our economy and a very complicated and important part of people’s lives, doesn’t always translate, right?

We know how to write code, and it’s not surprising that an AI can do a very good job of that. Navigating the complexities of the realities of healthcare and the realities of the fact that someone’s health and how they manage it depends on their biology, depends on their social circumstances, depends on all of these things coming together, and then they’re doing that in this really complicated system with arguably the most trained humans we as a society produce are those clinicians who are out there managing these. I mean, it’s amazing what they can do, but they’re doing it, as we talked about, in this fragmented, siloed way.

So, I think there are really exciting innovations that are sort of language-based, allowing clinicians to rapidly search and look up, in a language interface, clinical trials data, literature data, I mean, evidence for the literature, it’s phenomenal, it’s very helpful.

Something that Ted and I work on together, is trying to take that next step to understand, kind of, where does intelligence that you want to make artificial actually live in healthcare, and that the system as a whole is a form of intelligence, that a lot of the knowledge is in clinicians who’ve learned really how to treat real patients in the real world over time, and it’s not in the textbook, it’s inherently not in the textbook, and I think that’s a real limitation.

I think the really big problem, and we’re hearing it a lot from, you know, the C-suite of healthcare systems, is they have a ton of point solutions that are basically wrappers on language models to do some particular task, and the healthcare system’s so big and complex, it’s, sort of, this death by a thousand cuts. You know, I have a tool to help cardiology scheduling, that’s one tool, and then there’s another tool over here and another tool over here, and I have a revenue cycle management tool to optimize payments, and how well these work and where they work, and a lot of them signed up for these kinds of tools early and then are realizing that they’re not quite doing what they…

So, I think that’s where you’re getting a sort of frustration, and so I think what is exciting is that there’s a view that it’s going to take a little longer. It’s not like this magic bullet is going to show up and transform healthcare overnight, and that real integrated platform solutions are the kinds of things that can, kind of, work and actually even be more cohesive over the culture in that sense. We can come back to some of the challenges of that, but I think that’s where the lay of the land is today, at least in my view.

[00:16:33] Jennifer Chatman: Yeah, I just am reminiscing, Jon, about a talk that you gave to our golden grads a year or so ago. Our golden grads are Haas graduates who graduated 50 years or more ago, and you gave this great talk about the various use cases and the status of those use cases of AI in health, and you said, like, remote monitoring is doing great, and then you gave a case for complex diagnosis, which was not as far along, but then you gave the case of optimizing Medicare and health insurance, and that audience was on the edge of their seats. They could not be more interested in what you had to say. It was absolutely fascinating, and they were happy to hear that optimizing health insurance for individuals is actually something AI does pretty well already.

[00:17:26] Jon Kolstad: Exactly. I found my people. It is rare, I can kill a party fast with talking about Medicare enrollment, but it is true that I think, and this is kind of the core of it, that I think what you’ve had is a bunch of people, kind of, appropriately saw early versions of ChatGPT, saw what was going, and said, “These models are amazing, let’s bring them to healthcare.”

But the flip of that is we’ve had a really phenomenal ability to do a lot of AI, to do a lot of really good prediction, right? All these models are just prediction. It’s the next word prediction, it’s prediction of, kind of, what’ll happen to you next, but I think really getting to the core of what is the question that you can use AI for, and the complexity of healthcare is where there’s a really big opportunity. Health insurance is a great one, right?

When you need to choose health insurance, it’s actually about, for me specifically, what’s my next year going to look like, and how will that match to my needs? And so, that’s one that I’ve had the good fortune of really been working, both research and entrepreneurial work, in that space for some years now, and I’m really excited about it because, well, it does, you know, it hits the golden grads, but it’s a little boring for other folks.

But I think if you take a step back and say we can do all we want about taking your medications and getting healthy, but if you enrolled in a health plan that doesn’t even cover your medications, you’re spending thousands of dollars out of pocket, and you’re on a fixed income, right? It’s no amount of nudging and doing all these other things, and so I think health insurance really is an exciting opportunity to improve healthcare upfront.

[00:18:58] Jennifer Chatman: So, Ted, I mean, here’s the big question. Can AI play a genuine role in creating more coherence across all those fragmented actors and that siloed information in healthcare? Is that a realistic goal? Are you seeing it happen? What do you see as the key use cases in that sense?

[00:19:19] Ted Robertson: First, I’ll say Jon and I, and I think a lot of people in health AI, really take very seriously the potential downsides of AI, right? Like, “Is the security right? Is the privacy right? Is the transparency, right? Is it, you know, fair to all people? Does it actually work?” Like, we think through all that. I think, though, that ultimately, in healthcare in particular, we are optimistic, and I think public sentiment shows that where there’s some skepticism broadly in AI, and there’s skepticism, rightly, in the ways I just said, in healthcare, there’s also people sense some potential upside, and I think that includes culture.

I will lead with saying I think, over the next decade, it will change everything in healthcare. So, like it or not, I think it’s there. It’s going to change the basic data. It’s going to change every part of the system. It’s going to change how you touch it, and so we can lead and manage that or not. That’s a lot around culture building. So, just as a one big statement.

And then I’ll say in terms of culture building very specifically, this whole conversation around the gaps, the structure that’s in… The incentives are off, like maybe people broadly are mission aligned, like they want to help people and do right, but in the tactical, it’s off, that the decision-making with AI shifts from a bunch of individual experts to individuals still there, absolutely humans in the loop, but with this system collaboration, a broader set of insights and intelligence and collaboration, that’s the promise. That’s what we’re working on.

And I think that’s potentially going to shift a whole bunch in healthcare. So, the first successes of, like, sepsis alerts, great. They had to then figure out, like, “Oh, we need sepsis alerts in the workflow. Those are human decision-making questions.” Great. “Oh, now gen AI, we need to figure out chatbots and how that fits.” Like, “Okay, those are all one-offs. How do we now do this piece of this collaborative decision-making across the whole system?” And I think that’s one very big thing that’s going to change in the culture.

[00:21:25] Jennifer Chatman: I understand that you’ve been developing and deploying something called a large clinical behavioral model, LCBM. Most of our listeners have never heard that term. In plain language, how is it different from AI like ChatGPT?

[00:21:41] Jon Kolstad: It’s getting this collective knowledge is the way knowledge exists in healthcare, and it’s also experience.

And so, really, the LCBM, at its core, and this started as a research project about two and a half, three years ago here at Berkeley, and, you know, we subsequently deployed it and built it and spun it out, but really the core idea is that can you actually learn how a patient’s going to navigate through the healthcare system? That is a form of intelligence, that it’s not just a one-shot. What did the clinical trials say for this patient? It’s a patient comes in with some, kind of, unknown symptoms, you start doing some tests, maybe they give you the answer you expect, maybe they don’t, then you move on to some next ones, and then you do some treatment. That might be a, kind of, rapid experience, like an emergency room, or that might be over the very long run with managing a chronic illness. And in a chronic illness, you might be managing a patient that has six other comorbid conditions.

That path and navigation is the intelligence that we want to learn in healthcare, and so really the LCBM is this behavioral model. It learns how this behavior unfolds. I think it’s basically what we’ve been hearing increasingly talked about as a world model, that probably the most similar parallel, would be the kinds of AI that drive your Waymo, right?

Those aren’t trained on language, right? If a Waymo was learning from a language model, it would, sort of, you’d feed it the driver’s handbook, it would know all the rules of the road, but you would not want to get in that car. It would either stop because it’s following all the rules and nobody else is, or it would be unable to navigate.

Instead, how are those models trained? They’re trained by looking at billions of decisions by real drivers in response to what that car is, you know, quote, unquote, seeing, right? What the lidar, what the radar, what the cameras see. Would I turn the wheel left, would I turn the wheel right?

The LCBM, you really want to think about that as the healthcare equivalent. If a clinician knew everything about Ted up to this point, you know, his images, his labs, what might they do next? And then, if they do that next, how is their care going to unfold over time? And that turns out to be really powerful for, sort of, true clinical intelligence. It’s also really powerful, kind of, coming full circle that we talked about at the beginning, that starts to be a fabric that can basically pull together these different disparate aspects of the culture, right? So, if you think about being referred to a specialist, all of a sudden, that’s a problem of, okay, the GP or the primary care doctor here is saying, “I think you might need to see a cardiologist,” but they don’t really know what’s wrong.

Well, we’re going to have to wait to see that cardiologist for some, you know, six to nine months to even figure out what the next step is. How likely is that actually to be the right place to go, and what might unfold? That turns out to be able to bring that information closer in time, and so we’re really excited about that potential, kind of, back to this more integrated healthcare system that we’ve been talking about a lot in the background.

[00:24:39] Jennifer Chatman: So interesting, and that was a huge question of mine, was you couldn’t just be using language in the way that LLMs do. So, Ted, Jon started talking about how LCBMs address culture problems. Do you have anything to add there?

[00:24:56] Ted Robertson: A couple things I would throw out, Jenny. First, one of the things I want to underscore here is, right, it’s not text, it’s around human decision-making, and a little of what we’re trying to say is, like, what’s amazing about this, this isn’t a little step, this is a big jump to capturing the depth and width of collective human medical knowledge as it’s living and breathing, which is a lot of where it is.

So, you could learn the best practices for a low-income Black community someplace in the United States, that’s like they’ve really figured out how to deliver care there. And now you can translate that to many other places in a way that couldn’t be done or was done in snippets or very, with long lag times before. That’s a real change, or the best cardiac care to a remote place in Wyoming or Montana, right?

But it’s still capturing human decisions, real people, and that knowledge, and I think that just to, sort of, underscore that point from a performance, from a technology, but from a culture perspective, and what’s at the heart of that is still the best of humans, and I think it speaks to why people go into medicine and what we’re trying to do with healthcare.

I do think, in terms of culture and how it will start to affect things, is that it will reduce the need for people to always reconstruct context from one thing to the next. It will allow for creating more alignment of priorities, like, okay, this is what the right care with the right person at the right time, we can see that and actually predict, not just from a scheduling perspective, Jenny’s coming in at 2:30 and we should get to Jenny first, but that whoever is coming in, that they’re supposed to be coming in for that, and they should come in at month one, not month six. And that, sort of, triaging will happen much better than even the best of, say, the Kaisers of the world these days.

So, I think that’s a real change. I don’t think it gets rid of the complexity of healthcare, so I don’t think any of this should be read as, like, “Oh, suddenly it’s going to be a panacea or easy. There’s a lot of work to be done,” but I do think the coordination across care and the integration will happen better.

[00:27:06] Jon Kolstad: I think, just to layer on, I think it’s worth noting that right care, right place, right time, we’ve been saying that for like 50 years, and similar to, you know, we wanted AI to, like, be able to write a full story, right, and that is what this moment is an inflection, that the same architecture that allows ChatGPT to write a paragraph that has context now from two pages earlier, we now have the technology to have context for a patient that takes in some care and treatment they received three years ago.

So, the technology gets this core idea, which we’ve known to be true for a long time, but there just simply hasn’t been the level to have that kind of intelligence that’s there continuously.

I think a nice example is, you know, we’re running our models in the… we built one in the emergency room, and if you think of it, hopefully none of you had to go to the emergency room, but it’s a triage, it’s a diagnosis, and then you, sort of, need to make decide whether you’re admitting someone or not, and it’s often being done by a fragmented team. Maybe it’s being done, kind of, by an individual that has some latent belief about what’s going to happen, but they don’t know it yet.

And so, our LCBMs can show about 60% prior to the actual admission, our models can already say, “We know this person’s going to be admitted.” And so, that all of a sudden is powerful for the reasons Ted’s saying, which is you can get the right bed, you can get the right placement, so you, sort of, take this intelligence and start to sew together these pieces that have been fragmented because of that information flow.

[00:28:37] Jennifer Chatman: Yeah, I mean, what I’m hearing you say is that AI doesn’t just allow you to learn from the masses but it also focuses on the individual specific history and what that looks like relative to the population, which is, sort of, a matrixed approach to both generating knowledge, both from the power of a large sample size, but also customized to the specific patient. It sounds very powerful.

So, we always try to end the pod by asking our guests to give us, sort of, the two or three biggest takeaways, and so, in your case, I would ask, what are two or three ways that you’d like our listeners to understand how AI systems might genuinely improve the culture of healthcare organizations, not just efficiency?

[00:29:34] Jon Kolstad: Yeah, I mean, I think it’s about infusing intelligence into a system that is one of the most powerful things we’ve even built as a society. If you go back to Hippocrates, we have this system that can deliver amazing things to everybody, and it’s basically, right now, falling down because of the fact that there’s so many silos. There’s not a culture as an organization.

So, I think what’s really exciting to me about AI is not just operations, it’s the ability to actually have this navigation that will make each of these pieces more cohesive, and that’s going to be better for patients, but it’s also going to be better for those clinicians who are getting burned out, who are doing amazing work, but then basically, you know, they’re frustrated too when you’re telling them the story again.

So, I’m really excited about AI that is built for healthcare, that takes into account the realities, but also sees the opportunity for transformation, and I think that’s coming. It’s not tomorrow, but we’re going to start to see it more and more, and then I think, you know, it’s like when Hemingway said about going bankrupt, “slow, then fast,” and I think then we will really see that, sort of, those leaps that we can navigate.

[00:30:47] Ted Robertson: Jenny, if I could build off that, I think that’s a good place because I think, in many ways, patients and providers, all types of providers, anyone in the game, are victims of a just broken system that doesn’t make sense, right? And culturally, AI, by being this augmented, accurate intelligence, and I think part of what we’re saying is, like, the same way you think about the early days of the internet when there’s all this promise and then it, sort of, broke, it didn’t really pan out, but then it did come a little later.

We’re, sort of, at that point where everyone said big data, it’s all going to come, and it didn’t really work, but actually now it is starting to go, and if we do it right, it’s going to elevate patients and providers and let patients be more empowered, let providers play the top of their license and actually deliver and receive better care and be healthier both themselves and extend it to people who are not even fully getting as much care as they need.

So, the promise on lifting up those two groups is tremendous.

[00:31:46] Jennifer Chatman: Well, I have to say this has been a fascinating and uplifting conversation. It gives me great optimism and hope that you both are working on these immense challenges with incredible opportunity to really improve our health and help people get health delivery that works for them. Thank you both. Really enjoyed having the conversation.

[00:32:11] Ted Robertson: Thanks, Jenny.

[00:32:12] Jon Kolstad: Thanks so much for having us. It’s fantastic.

[00:32:17] Jennifer Chatman: Thanks for listening to The Culture Kit with Jenny & Sameer!

[00:32:20] Sameer Srivastava: The Culture Kit Podcast is a production of the Berkeley Center for Workplace Culture and Innovation at the Haas School of Business and is produced by University FM. If you enjoyed the show, be sure to hit that subscribe button, leave us a review, and share this episode online so others who have workplace culture questions can find us, too!

[00:32:41] Jennifer Chatman: I’m Jenny.

[00:32:42] Sameer Srivastava: And I’m Sameer.

[00:32:43] Jennifer Chatman: We’ll be back soon with more tools to help fix your work culture challenges.

Updated May 19, 2026

View posts by Laura Counts

Previous The Office Has to Earn It: How Physical Space Shapes Organizational Culture