Former FDA Regulator & Pharma AI Exec REVEALS: "Patients Trust AI More Than Doctors"

Speaker A: Healthcare and AI, you know, it's like if you do it wrong, you lose lives. If you do it right, you save lives. How do you actually innovate in an industry that has to be more cautionate than others?

Speaker B: Especially in healthcare, which is a very heavily regulated industry, you want to take safety and innovation together and you need to. I look at it as disciplined innovation. In the past, understanding how proteins fold, it used to take years of lab research to do it. Now we can do it in days, right. And perhaps in months.

Speaker A: Uh, you're educating me right now, now, so I love it. Keep going. How do you guys verify that there is a digital twin medical liaison AI company that is doing it right?

Speaker B: Consider these AI systems as highest level of scientific expertise. So now your doctor, he doesn't have to wait until the business hours to call up an MSR or another scientific expert to get that information. He could go, um, go to that AI model and get that information.

Speaker C: Please welcome Kausar Riaz Ahmed, the senior Director, Head of AI Precision Oncology at AH, AstraZeneca. A seasoned expert in regulatory affairs and computational modeling, Causar has spent her career at the intersection of science and global health strategy. Today, she's leading AI innovation for one of the world's largest healthcare companies. We are thrilled to have her here. Kausar, welcome to the show.

Speaker A: I've spent two decades guiding CEOs and founders to scale and reimagine what's possible. Now, the biggest shift of our lifetime is here.

Speaker D: AI.

Speaker A: I'm Seem Alexander, founder of Disruptive AI, co chair of DC Startup and Tech Week, and the host of the AI CEO podcast. If you're ready to learn how AI is transforming industries, creating new growth and rewriting the rules of business, this is the podcast for you. Because the truth is, the leaders who adapt now won't just survive, they'll define the future. Let's tap in. Quick note before we dive in. Since this episode was recorded, Causa Riaz Ahmad has moved from Pfizer to AstraZeneca, where she now is the senior director and head of AI in Precision oncology. Now let's get started. Welcome to today's episode of the AI CEO Podcast. I am your host, Seema Alexander, and I'm excited for season three. My second guest of the season, Khazariz Ahmed. Uh, she's the director of Generative AI for Medical engagement at Pfizer, which is one of the largest healthcare, uh, companies in the world. Right? Or healthcare pharma company. Yeah. Right. And, uh, again, as always, I Always share how I m meet my folks. And we did a amazing futurist series in partnership with John Hopkins at DC Startup and Tech Week last October. And we Happen to be LinkedIn friends or LinkedIn reach outs. Amazing. And that's the beauty of networking as we know it. But, man, when I hear your background. Nine years initially at the fda, heavily into regulations, now leading and directing one of the biggest practices in pharma for, uh, you know, for AI and all the things that you do. And you're also on the board to identify opportunities for AI startups that you're investing in through Pfizer's, uh, fund, right?

Speaker B: I'm not on the board per se, but I have been working with Pfizer Ventures internally in developing strategies for investments in AI native and AI augmented companies, and Phi Mine Healthcare.

Speaker A: So when I say board, I'm saying you are part of that committee. You are helping initiate a lot of the conversations and doing some of the analysis. So I think we're gonna have a very interesting conversation. Healthcare and AI is such a big topic. Um, and you have this, uh, really incredible balance between regulations and innovation. Right? That's a, that's a, that's, I think, a, uh, really cool background because that's one of the biggest challenges when it comes to, especially in healthcare and AI. So let's start there. First of all, welcome to the podcast.

Speaker B: Thank you.

Speaker A: Thank you.

Speaker B: It's such a pleasure being here, Seema. And, you know, I, I really resonate with what you said earlier about sometimes you don't realize how your paths cross. I think we had been connected on LinkedIn for a while before we got in touch about the, um, deceased Tara Beak. And I'm so glad to sit down here this afternoon and discuss AI and pharma with you.

Speaker A: I love it. I love it. So I had, um, Ainslie McLean here earlier, a couple probably, uh, episodes ago, and she was the former Chief AI Officer for Kaiser Permanente. So we went deep in healthcare, but more into precision medicine and things. I think we're going to touch on some of it. But one of the questions I have for you is healthcare and AI. It's like if you do it wrong, you lose lives. If you do it right, you save lives. What's your perspective on that? How do you actually innovate in an industry that has to be more cautionate than others?

Speaker B: That's such a fantastic question. And that pretty much sums up a good chunk of my career.

Speaker A: Is this a thumb?

Speaker B: Um, like you said, I spent about nine years at The FDA regulating products. And one thing that I realized was in healthcare you have to get it right every single time. Because what is oops when it comes to AI for some other industry might be a bad customer review. It might be a matter of losing service for a while. But here, an oops moment with AI or for any other, um, misstep could potentially mean delayed diagnosis for cancer. It could also mean, um, um, wrong drug dosage or it could potentially mean even the loss of a life. I think what we do here is a lot of times, especially in healthcare, which is a very heavily regulated industry, you want to take safety and innovation together and you need to. I look at it as discipline, innovation. So I think a lot of times you think of safety as your brake and innovation as your gas pedal. But I think in this particular instant you need to build innovation right from the get go. It needs to be guardrailed by regulatory, uh, frameworks. And I find that when you do that, I think what regulations actually help us do is allow you to get to your customers, which in our case is patients in a very. Deliver your end product, which is safe and effective medicines, uh, at the speed of clinical need. If you do it, if you were to take regulations as part of your foundation as you're building up your innovation.

Speaker A: M. So it's funny because, because uh, I have this ongoing meme in our house. Okay. So my kids are older now, 11, uh, and 16, but my older one and her cousins, when they're probably 10, would watch these silly pharma ads. And I say silly because they're all like, it's going to save your life and it's going to take away your migraine, it's going to do all this and then the other half of the whole commercial is. And you could die like or. And it was, it's hilarious when I showed you the skit. But my point is that there's always a balanced message when it comes to creating pills, creating anything. So does AI blur any of that? Like, you know, because. And uh, I'm just trying to play it out right, because there always needs to be, it sounds like there always needs to be a balance, even if there's a breakthrough.

Speaker B: Mhm.

Speaker A: So how does that work?

Speaker B: And your kids are very right to joke about it. I mean a lot of times you have the effect it is supposed to deliver and then you have all of those side effects that get mentioned in any of those adverts out there. But I think it's a fair requirement to have a balanced benefit risk Profile, uh, in all your pharma advertising. And I think there's good reason for doing it. I think AI can sometimes blur those lines because when you are putting out promotional material out there, I think AI by design is developed such that it focuses on the number of clicks that are generated. So if you were to imagine a situation where, say, for example, um, AI is an AI model that's working on developing content and if there are increased number of clicks on, say, um, on the efficacy of a drug, for an example, and there are limited number of clicks or there's limited interest or eyeballs or whatever your metric pressure. Yes. Whatever that might be on say, for example, the adverse effects of it, then in a very, um, unsupervised setting, what AI could potentially do is as it learns and relearns through the whole iterative process, it could focus more and more on your, uh, on the efficacy and try to, as it's developing the promotional material, somehow get all the adverse effects and the side effects sidelined.

Speaker A: So it's going to unbalance.

Speaker B: It is. And I think that is why it is very critical that you have your human clinical experts or scientific experts or what have you in the loop as it's going through those multiple iterations. AI can do the drafting, but then I think it is these humans who need to come in and make those decisions. As long as we have that, the skills are balanced in that manner, yeah, I think we should be fine.

Speaker A: Okay, so let's talk, let's go deep here, because how are you fine when companies like OpenAI are creating OpenAI health? But even before it, trust me when I tell you, I know so many people that are putting in health records, and I have a very concrete example for you in open air. Ah. In ChatGPT today. Right. And it's basically becoming an, a way to diagnose you and diagnose pills and kind of like one of the, I think one of the biggest problems or challenges, I think as a consumer, if you're taking multiple medications, you're like, all right, what's interacting with what? When should I take this? Is, am I even taking the right thing? I have multiple doctors telling me different things. Do they understand how this, all this stuff interacts? And now we have a system, um, trusted or not, that people are trusting, is that nothing? And actually in a lot of ways, they're getting diagnosed. Uh, so how does that work with what you are saying? Because what you're saying is, if I was able to control the situation, that's the ideal and I love what you said about AI can change the balance. Mhm. This is reality now. It's going to change the balance. So how does that work?

Speaker B: It certainly is, it is interesting. Healthcare is perhaps the biggest use case right now for AI. And um, the two big releases that we had earlier this year, cloud healthcare and GPT health, I think they have had such huge impacts. I mean there's something that I really love about it because number one, it empowers patients in a very substantial manner. Now they are able to. Patients are increasingly engaged with taking charge of their health data as compared to what they were doing in the past. Um, so there's an advantage there. The other thing is it is giving the power in the hands of the patients where they're able to educate themselves. So it's group health literacy and come down to the table when they're meeting with their physicians and have an engage and shared decision making. So there's a lot of patient empowerment that is happening. So I want to put that out first because I love what they are doing, what GPT has done with health, et cetera.

Speaker A: So pharma loves that or is it you loving that? You know, I'm just curious. It's a question, right?

Speaker B: I think it's a mixed, it's a mixed pack because I want to come back and address what you said, which is about whether you want it or not, in many ways, regardless of how you frame it, people are getting diagnosed, people are treating the advice or whatever the output that's coming from these platforms as diagnostic advice. In that context, I think it's a mixed bag for me and I think perhaps for a lot of people in pharma as well, because we know this is the direction the future is going. But I think what is missing is the guardrails, for example, for someone looking at GPT Health and if they have their lab results and what they're trying to figure out is, oh, I see these multiple numbers on my cholesterol panel, uh, why do I see these multiple numbers? Or they have this disjointed data, for example, from the last three, four years and they want to see the trend. I think that is education. Yes, but, but I think what is happening is it goes beyond that, which is where, um, you know, someone might go feed their lab results into GPT Health and then they might in fact end up missing the next doctor's appointment, they might want to change the medication, they might want to change their drug dosage, etc. I think at that level there are disclaimers of course, that Claude and OpenAI have put forward. But I think having worked at the FDA for so many years, FDA does its review of products based on intended use. And the way these platforms are being used is for diagnostic recommendation. Obviously, you have all of those disclaimers. So, uh, from that point of view, I would certainly say that these tools obviously are not going to go away, but I think we need to have guardrails. And I think AI should be trained in a manner where it's not giving out, um, routine answers to every single query. But what it is doing is it needs to know when it's not sure or when there are complicated questions. It needs to be able to go back and say that it's not to redirect the consumer back to their physician, back to their clinician, or have language in there that after every output that it gives, there's a three, there's a.

Speaker A: There's a call to action. Yes, I think, and I'll be honest, like, um, I just did something with my cousin. Uh, unfortunately, she has chronic fatigue syndrome. It's my cousin. Yeah, It's a long story. But. So she's been, you know, sick for a bit, and we had a moment, um, just kind of going through some labs and doing some diagnostics, and it showed, you know, her. Her blood was a little sticky.

Speaker B: Uh-huh.

Speaker A: Which meant because I'm on blood thinner. So, like, I can say it where I was like, man, I think you got to get on blood thinners. Like, said, there's something going on. And the next time she went to the doctor, it was absolutely right. Like, you know, and to your point, they're like, first, you know, for her, you know, she was a little freaked out because she's like, oh, God, you know, and like, what does that even mean? But I was like, you know, like. But at least, you know, like, you know, and, you know, and it's. I think it's going to get stronger and stronger, um, in terms of that. And I'm curious how pharma is going to play. I think you have to kind of almost partner and be part of that process, not to drive the narrative, but to add to the narrative, to the balanced approach that you guys are seeking. Because I think it's going to take a mind of its own. I just don't see it being regulated in the way that you guys are

Speaker B: probably hoping, I think right now, because these, at, ah, the minimum they need to have. They need to be HIPAA compliant. I think what happens is when a patient say, for example, if I'm a patient and I go there and I put, voluntarily put in my information over there, the protection that I have from HIPAA disclosures is now gone.

Speaker A: Yeah.

Speaker B: And it's fair game at that point. I think pharma does not have a central regulatory role there. It comes from the FDA or other HIPAA compliance etc. But I think we do realize that you know, when we are engaging um, right now, the way the structure is built and I think for good reason, pharma directly engages, engages with HCPs, healthcare providers very in a very limited fashion with patients. I think you know what really the, the direction that pharma is going towards is being able to use digital platforms and digital engagement to get our message in a better fashion through various channels. We are now talking about omnichannel, um, engagement where you use social media, digital in person formats, et cetera to engage with the HCPs. But I think pharma does not have a direct role where it is meeting with the patients on a very routine basis as we would with the HCPs, but I think that does not um, so I think in this scenario where we are. I think a lot of the onus of this whole situation is with the patients. I think that's where. And I think the regulatory um, agencies, they are playing a little bit of catch up. So I think that's where we are right now. Yeah, yeah.

Speaker A: So are the pharma companies not allowed to speak to direct to consumer or direct to patient to educate?

Speaker B: You could have dedicated small um, group discussions or other types of. There is a limited window within which you are able to engage for education purposes. But I think for something like this I don't think pharma is able to directly engage with patients per se. M. So um, I think the one body who could do that engagement is MTE or other um, FTC and other types of agencies which have that um, onus as well as the power to come up with guidelines and guidances. Um, that would, that would make um, OpenAI Cloud and other such companies. I don't, I'm kind of hesitant to use this analogy but to, but to stay within those, within that fence or within that guardrails. I think they, I'm not sure if they realized, I mean obviously OpenAI had the data where they knew that over 40 million people were using ChatGPT for health purposes when we developed GPT Health. And I think, I'm not sure if they realize that these disclaimers are not going to be enough I think, um, the usage has quadrupled. It's increased, substantially increased.

Speaker A: Because people, once they experience something, I say this at every podcast, but there are aha, oh, shit moments like, did that just happen? Can I actually understand my health care now? Like, uh, what do these blood results mean? Or what pills I'm taking? So the more they trust, the more that they're going to use because it's convenience, it's saving them time, potentially saving them money because they're like, oh, wait, what should I be taking now? My question back to you is HIPAA compliance is the big thing. We sign it every time we go

Speaker B: to a doctor office, right?

Speaker A: And it's like, we want to keep your data for your health care private. It's just been a thing, right? Because like consumers, patients, that's an important piece, right? Uh, just overall in terms of a. Systematically. So now we have people volunteeringly putting their information, probably not even redacted, meaning their names are not taken off or birth dates and information. So what does that mean? If I'm a consumer and I choose to put my information in ChatGPT, my data, my healthcare information, to get a prognosis, to get help and support me, what are the pros and what are the cons?

Speaker B: Well, I think when you put your data out there, I think the pros are you're able to. Depending upon. Now I'll have to take a step back here and say, even, uh, as I define the pros, I'm looking at GPT as well as Claude as health literacy tools, tools that provide you health information, not necessarily diagnostic tools, but obviously that is happening, that's what's happening right now. So I think what the pros are, that it promotes health literacy, it empowers patients. So for example, they have a doctor's appointment tomorrow morning, they put in this, they are able to prepare well for that engagement, they are able to participate better and um, have a shared decision making about their health decisions. So those are certain things that are pivotal for patients that are not there in the past. Um, but I think the cons are a lot of these companies, they have disclosures out there that are liability protections for the companies, not necessarily offering patient protections. What they are telling you is, well, we told you, don't trust us. But then the platforms are designed in a manner where they encourage patients to trust them or users to trust them.

Speaker A: That's right.

Speaker B: So I think in some ways it defeats the purpose and I think that's a huge danger zone that users are getting into. And I think A lot of people perhaps don't realize that when they put their information out there. Although, although OpenAI and Claude, um, they have been very vocal about the fact that the data that patients put out there is not going to be used for training purposes. But then you never know when the terms and conditions are changed. This could change in six months, those could change in three months, et cetera. So I think there's certain amount of risk that, that individuals are putting out there by, by sharing the information uh, with these platforms and I think those are huge risks and I think the regulatory landscape as such, because the innovation, the pace of innovation is so fast, um, that health regulators are essentially playing catch up to a certain extent.

Speaker A: I think all regulators playing catch up across the globe. Across the globe, across industry. It's a lot to take in. Except if you listen to the AI CEO podcast and things that help, help bring some layman understanding to what's happening in this world. Right, um, no, I couldn't agree with you more. I just think that um, the pace to your point is moving way faster than we think and consumers uh, are going to do. It's almost like they're going to do it anyway. And I'm just curious of OpenAI or one of these platforms will get sued eventually because to your point, I just wanted to address this. You mentioned. Well, they say it actually doesn't get shared in the model or like added train, but people don't realize until you turn that setting off and in data controls because everything will get shared in the model unless you put in like if you go to data controls in OpenAI or Perplexity or any of these LLMs, you have to physically say don't train the model with my information. And who knows like you know if that's even true. Right. Um, just curious though. There's going to be a lot of micro language models, right. Which means these are, you know the, the chat GPTs and the uh, Gemini's are all these large language models, the micro ones that are going to be focused on healthcare.

Speaker B: Yes.

Speaker A: Right. That are going to be very specific to ensuring and I'm sure GPT Health and others have done in some capacity but like very. All the data, all the records, all the research, all the things. So that. And potentially going even deeper and securing people's data. That is the future. And I've actually seen companies on the startup space that are already preaching that we have a thousand doctors information already in here. This is like free doctor care. That's what they're saying now. How Is that possible in a world of regulation?

Speaker B: I think, um, again, a lot of. So with this startup scenario, um, whenever any of these tools offer you any kind of a diagnostic information or they offer you information that supports clinical decision, they need to be rigorously validated. I think that's a, uh, knowledge gap. That's also a gap because a lot of these startups, they are in the initial stages and I think many of them have not partnered yet with either say, an academic institution or a health system like Memorial Sloan Kettering or MD Anderson, uh, or Johns Hopkins, what have you. So I think a lot of times, I'm not sure if these startups are even realizing it, you know, having either SOC2 compliance or HIPAA compliance or software as a medical device or a clinical decision support software, um, validation. I think those are essential, um, needs

Speaker A: now in the infrastructure itself. So if you're building anything for healthcare or pharma. Pharma. Uh, those are foundational things that you have to incorporate. Absolutely.

Speaker B: Those are. That need to have. Yeah, that need to happen on day one, even as you're planning your tool. I think a lot of times what is happening is that these are, these are considered an afterthought. And I think that's when you run into issues. Because now you have a really good model, one of your micro models, but then you are not able to leverage that. You are not able to get, say, a health system or a pharma company or any other, uh, customer for that matter, to use it because there are restrictions out there.

Speaker A: They're not enterprise ready.

Speaker B: They are not enterprise. Yeah, they're not ready yet. We think micro models are not yet.

Speaker A: Yes.

Speaker B: And the micro models are great though, because I think I look at GPT and OpenAI and Claude and Gemini and others as your general practitioners. Right. And then you have these micro models which are like, say, for example, a specialist and oncologist who's spent his or her entire life looking at that specific data. So I think micro models offer a lot of advantages. I feel like if you were to build them responsibly, um, because micro models are trained on such specific data. So the hallucinations, the rate of hallucinations, which is essentially inaccuracies in data, um, that is much lower. That's what we have seen from publications, uh, the audit trail can be established very easily. And I think because these are smaller models, they can perform faster as well. There are so many advantages to it, I think, because of not being able to plan better. Um, I think that's why I Feel like I'm saying m this over and over again. Where you want to have in healthcare, you want to keep your safety. By safety I talk about regulation, um, I talk about having those governance guardrails, et cetera. You need to always keep them together and a lot of times that happens at a later point and then you realize that's when you have that oops moment and then you come back and you try to fix things.

Speaker A: Right. What is the governance model like? Give me a couple core, uh, sort of focuses on that. So if I was a startup in the space I would know that these are, or is it or what you've already shared with us.

Speaker B: Um, for example, when you have, say for example, you're building a micro model again and the data, the training data that's used to build that model, having data equity in there especially say, for example, I'm going to use um, a micro model that can be used for example to um, diagnose or even to help you with prognosis, et cetera, or even to provide that kind of information. When you have data bias in the underlying training data set, I think it gets exacerbated, um, especially if there's not enough representation of say underserved communities, uh, you are not able to um. I think at the beginning of the talk we mentioned personalized medicine, precision medicine, being able to um, prescribe the right medicine to the right individual. That's not going to be possible if you are depending on these type of models for your information gathering. And the models do not have that type of training data set that gives them the ability or the power to make the right recommendations or give you the right outputs. So I think governance models need to consider things like data equity and data bias, but also transparency, uh, having that clear audit trail. Mhm. Uh, explainability. Right. And I also think governance, um, when you think about governance models, you think about um, responsible deployment as well, which is also a piece of it. Um, I think with AI adoption and with these AI models, one of the biggest challenges these days is um, change management. How do you drive change management in a responsible fashion?

Speaker A: Yeah.

Speaker B: So those are a few things to consider as well as you're building your governance structure.

Speaker A: Yeah. So I'm um, going to switch the conversation to. I mean pharma, uh, also has some big challenges.

Speaker B: Right.

Speaker A: Just in any industry and how you guys are looking and framing AI as an opportunity. Right. One of the biggest, I would say patient, uh, challenges is the cost of drugs.

Speaker C: Right.

Speaker A: And uh, I think I shared with you during a prep I was at a John Hopkins event on AI and innovation in health care recently. Mark Cuban was there and he, he was on a very spicy panel with Dr. Conway, was the CEO of Optimum. Right. And man, oh man, because, you know, I think everybody knows who Mark Cuban is. Serial entrepreneur, but now he's also a CEO of Cost plus, which is like a, um, you know, a drug, uh, it's a drug company that's creating the opportunity for distribution at a lower cost, direct to patient for certain things. And he, he says he's blocked in certain ways because healthcare systems are very complex. And he didn't say it in the nice way. He did it in the more human way. And he went in on this Dr. Conway, like I've never seen anybody go in on somebody on a panel. And part of it is he has a lot of conviction as an entrepreneur of like, how do we make this system better? How do we make the drugs che for the people who need it most? Right. And so what's your perspective? And I know you can't talk about it from a Pfizer perspective, but just in general, your perspective on how can AI help, uh, with the advancement and rapid development of drugs to the extent that the cost gets lower to the patient?

Speaker B: I think that's a very nuanced question.

Speaker A: It is, but it's a challenge, right? It is a challenge.

Speaker B: So I'll give you a few examples.

Speaker A: Sure.

Speaker B: Say, for example, um, you have an AI model that can, um, predict if a drug, um, that's going through clinical trials, um, going through clinical development is likely to fail. And if it's able to go do that with meeting the benchmarks for accuracy, then you're probably going to end up saving hundreds of millions of dollars. Because for a phase 3 clinical trial could sometimes go over the cost of a phase three clinical trial could sometimes go over 500 million. So I think phase three clinical trial

Speaker A: could go over 500 million and it

Speaker B: could last multiple years. There are trials that run beyond four, five years as well.

Speaker A: So you guys are just trying to recoup your money, is that what I'm hearing?

Speaker B: Well, but AI, the thing to consider is it could help you there. But then the other aspect of it is the cost of building those data infrastructures, AI model infrastructures, building the AI talent that allows you to make that prediction in a, uh, trustworthy manner, in a manner that's accurate, it is complete, and uh, something that can be verified. All of that also gets expensive. So I think when you look at drug development and the costs are Shifting, it is hard to say. The costs are completely going away. Say for example in R and D in the past, it would take years as a drug go through identification of a lead molecule, lead optimization, all the preclinical testing and then there's clinical testing that could go for years and then you have all the commercial activities and what have you. So yeah, I could shrink that timeline. It could shrink some of that timeline. Uh, obviously it can shrink, um, the number of years the clinical trial would last. Um, but it could help in R and D, it could reduce about 30 to 40% of the time. And cost, potentially that, that is huge. But I think the costs are also shifting like I said, because AI isn't all the supply chain infrastructure is not cheap at all.

Speaker A: So I think eventually there'll be economies of scale and as people build it. But like even with COVID they said the reason that we were able to get it to market was because of AI.

Speaker B: Yes, AI was a huge driver. We were able to get it dropped to the market in two years. That would have, that would not have happened without all these emerging technologies.

Speaker A: Right?

Speaker B: Yeah, yeah, so, so there's a lot of advantage there. And I think it's important to me at least to consider cost is one thing and that's the reality of the world that we are in. But it's also essential to consider, okay, uh, maybe it gets a little more expensive, but if you are able to bring a life saving drug to a patient much sooner, I think it's really hard to quantify that cost. I know economists quantify it, right? But, but, but I think, you know, there's that balance that we need to strike.

Speaker A: Yeah, yeah.

Speaker D: Before we get back to the episode, I want to say something very important. We're in a three year window where businesses will either rewire or get left behind. And that is what I call a yes and moment. Yes, AI is going to disrupt how things work. You hear episode after episode on this podcast. And the leaders who really understand it are going to be able to create massive advantages for their businesses, not just incremental change. But that's exactly why we created AI BusinessCon. It's not another AI conference. It's for non technical CEOs and business leaders who want to rethink how their businesses operate, reimagine what's possible and actually rewire how they compete. Now if you want to be part of that 5% of leaders who understand this shift, to take advantage of this shift, to take action, go to BusinessCon AI today and sign up I'm really

Speaker A: looking forward to seeing you there. What about clinical trials? Right. Like, what I'm trying to get at is like, okay, now there's a huge populations of, um, people who have different diseases, different chronic issues, and one they don't know about. Some of these clinical trials that are happening to, uh, they don't know if they qualify. Right. And um, you know, and it's like speed to communication at that point is really important for certain patients and their caregivers. How do you see AI supporting and helping in that space?

Speaker B: I think, um, again, I would want to go back to health literacy. Number, um, one is providing the health information out there so they are able to learn more about what they're going through. They're able to ask, ask the right questions to their practitioners. So that's, that's a great way. The other is being able to, by engaging by pharma, engaging proactively with the healthcare providers at their peace, um, by being able to, you know, this is a very interesting time and it's also a very fluid time because preferences are changing for everybody. I mean, ten years ago I used to read newspapers, but I haven't touched a newspaper in the last five, six years or even more. I think that's true for all of us.

Speaker A: I'm an audible podcast girl.

Speaker B: Uh, I read books and these days it's been more than a year since I have read a book, a hardcover. I listen to audiobooks, um, and other forms of media. And I think it's the same with Etsy piece as well. So I think where AI can really help is number one, understanding what are the behavioral shifts in HCPs and for

Speaker A: people who don't know or health care professionals or doctors.

Speaker B: Yes.

Speaker A: Specialized doctors. Regular doctors. Yeah. And depending on the drug, they need to have a better understanding of what any of the pharma companies are bringing out that actually works so that they can actually prescribe them to their patient.

Speaker B: True.

Speaker A: In a responsible way. Yeah.

Speaker B: Engaging with them, educating them about a new drug and, um, educating them about emerging science or emerging, uh, medicines or technologies that are becoming available. I think that's another way that I can help you because AI can help you understand what is the preference of a given doctor. Uh, is this doctor an early adopter of new medicines or are they lead followers? Um, who should I engage with in order to get my medicine to the patients faster? So, um, those are a few things. Those are other ways that AI can be helpful even when we can't directly meet with the patient.

Speaker A: That's a regulatory thing that you can't meet. Ah, like it almost feels. I mean, I get it. Right, because there's when, when pharma does it wrong, they do it wrong. Right. But I almost feel like, man, you guys are the experts. Like, why wouldn't you tell us? You know, like, uh, you know, and I, I understand that it's so regulatory wise. I'm curious because again, you have both sides. Does it make sense for pharma to create a database that has all the information, all the research, all the things, and then convert it into fifth grade language so people like, you know, like patients can understand it. Like they just. Or it feeds into chat, GPT or GPT Health or you know, whatever it is. Right. Because you guys are the subject matter experts. But at the same time there has to be a balance because that's not. You, uh, know, I, I know there's this. So let me speak to that. Yeah, yes.

Speaker B: Pharma is doing it though. So I think what I, uh, maybe I should clarify this. Yeah, Pharma can engage with patients. When you are educating the patients, I think what you do not want is a perception that pharma is trying to influence them to say, for example, purchase their brand of drug or another one. Yes.

Speaker A: So this GLP1 or this, whatever the T1 is.

Speaker B: So I think almost all pharma companies, um, be it Pfizer, Sanofi, AstraZeneca, BMS, um Takeda, any of those, every time a drug, new drug launches into the market, um, they do put out information about the drug, um, in layman terms so patients and the general public can learn more about it. So that type of engagement is happening right now. Um, a number of pharma companies have also started up. For example, Pfizer for all. Um, it's a website that's dedicated to patients, that allows patients to come in there and learn more about, um, the deceased states, learn about various medications that are out there and it is done with the purpose of educating them. Similarly, other pharma companies have also, Eli Lilly and others have also taken similar steps and they are doing it. But I think what, um, your regulators don't want you to do is influence, uh, them one way or the other to purchase your brand of medication.

Speaker A: Well, you guys normally go through key opinion leaders. So people that are thought leaders in the Dr. Space who believe we've used the drugs and then they can be honest. Right. Like that's just.

Speaker B: Yes, so we do that as well. So yeah, um, we engage with healthcare providers, key opinion leaders. Um, there's also patient education that happens. Um, and there is, um, Pharma has dedicated medical science liaisons, um, you know, who are experts in there. They are either PhDs, PharmDs, sometimes MDs as well. Their role day in and day out is to act as scientific experts who can share that information with your doctors. So that gets passed on.

Speaker A: Oh, my friend Ritu is the Botox medical. So now that, now you're saying that I'm like, wow, she used to go educate all the doctors and all the different use cases of Botox. And she's like, seema, it's not just for beauty. You know, you can use it for TMJ and you could use it for this. And I'm like, but that was her job, Right? I know. And she would pinch herself because she's like, this is an incredible job because I could really get to teach these doctors like all the realities and how we can do much more than what they think. Right. Um, so I, and I think you and I chatted during our prep that that is an AI opportunity and that there are certain companies out there creating digital twins of medical liaisons. So maybe speak to that and what that looks like. Because this is the mind shift I want people to have like, look, this is the current roles and this is what the process is. But how do you use AI, not just as this chatbot or as, how do you use it as a source of intelligence. Right. And I feel like you gave me a great example so maybe share a little bit more.

Speaker B: I like that you use the word intelligence, um, because when you say intelligence it means, uh, to me that is something that is, there's scope for learning, there's scope for growth there as opposed to just sharing information. Yet we are now sharing intelligence.

Speaker A: That's right.

Speaker B: Um, so think of a traditional medical science liaison. Msls. Um, they are, like I said, they are scientific experts. Um, they have knowledge in a given area. Your friend Ritu, um, should be a scientific expert with Botox. Know, uh, everything about all the medical conditions with Botox injections. Could be, um, life changing, could be helpful.

Speaker A: That's right.

Speaker B: But imagine a doctor who um, at 2:00am, um, needs to dose a patient and he's not sure what dosage fits that particular population or that particular. For, uh, example, the age group. Based on the age group, um, of the patient, maybe he or she is a child. So in those type of scenarios, I think this is where digital twins come in. These are AI models, um, that have access to the same knowledge as these human Scientific experts consider these AI systems as experts, um, that have expertise, the highest level of scientific expertise on almost all topics related to that particular, say for example a, ah, type of cancer, et cetera. So now your doctor, he doesn't have to wait until the business hours to call up an MSL or another scientific expert to get that information. He could go, um, go to that AI model and get that information at

Speaker A: 2am in the morning, because that's what they do. So actually, let me ask you that then, right? The reality is the model is only as good as the data. Bad data in, bad data out. How do you guys verify that there is a digital twin medical liaison AI company that is doing it right, that's doing it accurately and if there was ever any information that was incorrect, who is liable?

Speaker B: I think when we are validating them, the company, the vendor is liable, of course. So I think there's a lot of technical due diligence that would go into it. We would have to evaluate it with all the, evaluate all the training data that was used in building that specific digital twin. Um, like I said before, understanding uh, what type of training sets were used, data bias, data equity, et cetera. If you had, for example, if you had a digital twin that was created, you could use historical data and feed that historical um, data onto the digital twin and see if it's able to make the right predictions or give out the right recommendations. It allows you to go back and cross check because in the past you have used that historical data to make recommendations. So those are some type of evaluations that we would do to ensure that it's working out correctly. But I think once a pharma company has say, for example partnered with this digital twin vendor, I think at that point of time the liability suddenly shifts hugely to the pharma.

Speaker A: Yeah.

Speaker B: And then I think that is why a lot of these systems um, are slow in adoption. Because there's, if you want to do it right, the amount of trust that you have in humans, it's going to take a while before that trust can be shifted over to AI.

Speaker A: Um, and I agree and disagree. It's almost like I preach this stuff every day, right. I've been doing it for three years when I got fully exposed and was like, holy crap, this is going to change the world. And that's my platform, my mission in a lot of things that we're building and educating. Um, but I also feel like when people have the moment to like, oh my God, it actually can do this, it starts to switch. It's Just having those moments. Right? It's having those moments. But here's the question to you, because doctors aren't always right. Right. Even Data centers are 99.9% uptime. Right. There's always a 1 or 0.1% or something. Right. And they have liability. So why do we have to feel like the AI has to be right all the time?

Speaker B: That's a good question. Because it's not a ride or die situation. We treat it as a ride or die situation. I think what it actually is is we are, um, benchmarking AI to perfection. We are benchmarking it to 100% accuracy every single time. And that is important in healthcare, especially because how consequential a, uh, wrong decision can be. But I think what we need to compare it is with the current status quo. And if you have data that says Doctors are correct 80% of the time, if AI can be correct 90% of the time, 85% of the time, then I think, in my opinion, in my honest assessment, I think that should be enough to get us started. One thing to remember is there is a danger or there is a consequence to inaction as well. Just as taking an action, there's going to be a consequence, good or bad, when you don't take an action, I think the consequences also matter. You could have used an AI model to, um, improve or to better inform your healthcare provider who could have done a better diagnosis or provided better treatment options to his patients. Now, if you were to not use it because you're paralyzed by fear, then there are going to be consequences too.

Speaker A: Yeah, no, I love that. But here's a question. How much of the leadership of the pharma industry believes that? How much do they feel, um, that they need to be AI forward right at this stage, or is there still a lot of educational fluency in the industry that needs to happen so people understand your point?

Speaker B: Exactly. So, you know, in the last year or so, almost all farmers have put out information on their annual reports, press releases, et cetera, where they talk about how AI has improved efficiency. So a lot of times the conversation has been about efficiency. What it took us, say, days to do, now we can get it done in 10 minutes. So those have been the results that are provided out there. Um, I think improving patient outcomes, that's a higher bar. So that's coming. Um, I think leaders in pharma are very, they realize the potential that AI has. Um, they also realize that it's going to take a while. They see where their loopholes could be so there is cautious optimism, at least that's what I'm seeing here. Um, but they are totally bought into the potential of AI. Um, they realize that it's not a plug and play tool. It's going to take time, it's going to take optimization. I think that is where we are. I think we are in a really good phase where we are building out these models, we are building out these pipelines, but it is going to take us maybe a couple of years or um, what have you to actually see those results, to actually see improvements in Asian.com.

Speaker A: so you, what you mentioned was, you know, everyone's focused on the operational efficiency. One of the things I preach is it's. Yes, and.

Speaker B: Uh-huh.

Speaker A: AI is bringing a whole new rewiring of the way businesses run, the way intelligence layers are put on top of workflows and optimizing your workflows. And it's changing, just like the Internet. But 10x. Right. This is my wholehearted belief and I know it's happening. Right? So. And you know what's happening, right? So. And I feel like pharma companies has so much data, so much unique data from things that work, things that never did, things that were close, and add other additional data from the right data sources that can be a, um, very powerful, just data set for multiple things. Like we were so close at this cancer, you know, drug, but we were missing this. But now there's this new research that came in. Hey, what if we bring all this together? Like, look what it is. That's the beauty of intelligence layers. Is that where the industry is going or is that too far, too far fetched or too far forward?

Speaker B: I think the industry is certainly considering it, but I think what we described right now can be a very. Sometimes it could be a very academic question that you are trying to, in a say, for example, depending upon where a given pharma company is, the situation it is in. It could be. I think what you want to do with AI right now is to address immediate efficiencies, like you said, operational efficiencies. So this has been my understanding. I think that is what we are trying to prioritize because we need to demonstrate early wins. Yes, to demonstrate the potential of AI. And I think as you were starting the conversation, we talked about starting off with simple, keeping it simple in the beginning and then building up on it some building blocks. I think that's what bigger enterprises and pharma companies are doing as well, at least from my vantage point.

Speaker A: Um, but you need to bring me in as your keynote and so I can change the mindset and here's why. And you know, I do a lot of talks on why AI is bigger than the advent of the Internet. I go in specifically in different industries. So it was interesting for me. A couple of weeks ago, I got to be in front of 125 senior leaders in commercial real estate. Right. All different types. It's a very interconnected industry. And the head of the World bank, commercial real estate for their center of excellence was there. Um, and we had multiple leads after the conversation, but he caught like we chatted the other day and he said, listen, I need you to come in and debrief my executive team, because I've been looking at this incrementally, right. And we're responsible for a lot of tools and all the great things. But I want you to share what you mean by this intelligent operating system. I want you to share like, you know, and it's like that's the shift. I think it's a yes and right. It's a yes and it is an

Speaker B: S. And I think because of the competing budgets and other priorities that we are in, I think people are cautiously optimistic with AI, but they do realize this is where the future is going. I think across all companies. If you look at the annual goals, the strategic priorities that have been put out for say 20, 26 and coming years, you suddenly see an AI component out there. Um, I think there are certain projects that would be considered strategic moonshots with AI, and I think those are being tackled in a very, um, how do I say this? In a very. I don't want to use the word reticent, but they're very cautious in doing it with guardrails. With guardrails, yes. Um, and then again there are sandbox. Yes, sandbox environments. And there are certain other projects where you see immediate value, where you can see these are projects where they could easily go from pilots to um, a production grade AI system in less than six months. I think there's more prioritization of that. But I think in certain other topics, especially when it comes to say, medical affairs, engaging omnichannel engagement. That has been the buzzword with AI in the last two years. I think more and more AI is being used for those, uh, for those ends.

Speaker A: And just to understand omnichannel channel engagement is just marketing in different, multiple channels. Right. Like what's the aha, uh, moment for pharma, uh, for AI usage in those channels? Like I'm a little. Yeah, I'd Love to hear that.

Speaker B: Absolutely. So omnichannel is slightly different from multichannel. Multichannel is where you use multiple different channels. Omnichannel is understanding for this particular.

Speaker A: Personalized. Personalized. I got you. Yes. Sorry. I love it.

Speaker D: My bad.

Speaker A: But that's perfect. Right. So everything is about personalization and precision. So you're going where the people are at. That's what you want?

Speaker B: Yes, that's where you want to go.

Speaker A: Uh, where do the HCP sit? I'm curious, where do they. Healthcare. Because, doctor, one of the biggest challenges for pharma is educating doctors. Right? And because they're busy doing. They're operating or they're. And I say operating and meaning in their own, you know, uh, practices or operating like just being a doctor. And so it's hard to get in front of busy people. Right. And then especially when they're used to something that's already working, et cetera, et cetera. So where, where do you see them hanging out? Like where, you know, how do you enhance the education even in an omnichannel? Like, where do you see them hanging out?

Speaker B: So I think there's, there's been a clear shift in preference with them for a lot, a lot of doctors where they're going from print material or in person meetups, which like you said, because they have so many things that they're working on, they don't have the time to meet with individual medical Science Liaisons dedicated 30 minutes, an hour every day, that's not happening. So I think being able to meet them on social media, um, being able to meet them at conferences. So understanding what conferences do they go to, uh, what type of topics do they engage in, what topics do they publish, are they engaging with emerging technology, etc. So being able to gather that intelligence and when you interact them with, interact with them at conferences, bringing out, uh, developing your content, your engagement material in a manner that resonates with them. So I think that's what a lot of work that's being done in say omnichannel engagement or better driving an impact through these engagements and education of um, healthcare professionals. So that's what's happening.

Speaker A: Is that an ideal, uh, pain point for like a third party marketer, for example, working with pharma. Ah. Like, you know, is that something you guys are seeking or you're doing internally?

Speaker B: I think we are doing it both ways. Um, I think in. Sometimes I think you need a competitive differentiator. So you talked about data a few minutes ago. So when you have data that is unique to Your pharma and you wouldn't want to share it with others. And that's when we are building internal models. Ah, Models that allow us to predict, um, this doctor, this specialty in this geographical area should be engaged through this avenue.

Speaker A: Mhm.

Speaker B: Um, that could be LinkedIn. Mhm. Um, that could be docs and that could be other channels like opening.

Speaker A: And that's something farmers are creating today.

Speaker B: Farmers are creating today. That is a huge focus for pharma right now.

Speaker A: Yeah.

Speaker B: So that's what we are doing. I think there are also other, a lot of startups that are developing in that space. Um, I think there are startups and marketing, um, um, agencies and companies, um, that are either building out, um, we talked about digital twins in the past. Um, they are building out. You asked me, where do doctors hang out? Uh, virtual advisory boards is one mechanism to get a lot of doctors to hang out in the digital age. So there are companies that are building out these virtual advisory boards with AI that allows these doctors to hang out with like minded peers without having to travel, um, exchange that information, have that intellectual conversation. But, but do it in your own, at the comfort of your office, right? Yeah.

Speaker A: And then that feedback gets back to the pharma company as a report. So actually one of the questions I have, because I always figured getting on demand feedback from doctors, right. Like, hey, I tried this new protocol, like, you know, and it worked for this patient. It did not work for this patient. Here's what happened. Here are the symptoms. It could be anonymous, but like having some sort of Persona about the actual patient would be helpful, right? Because you know, do you guys do that today? Because you would think, right? That's the feedback loop. AI is all about feedback loops, right. And better to get the feedback loop from, but the person who's dealing with the patients or the patients themselves, if I'm being honest.

Speaker B: Yes. So one of the things that all pharma companies do is have patient journeys mapped out. So starting from your initial doctor visit to your diagnosis to say, for example, if you had surgery or at home care, and going through that entire map of it.

Speaker A: Uh-huh.

Speaker B: And um, developing that maps these patient journeys for different patients, accounting for the differences in the patients, be it demographic differences or gender differences or uh, urban versus rural communities, et cetera, that is something that pharma companies have been doing for the longest time. But now I think with AI you're able to do it in a dynamic fashion. So you are able to be more agile. And you are also. Because previously we had all of this information available, but it was disjointed and a lot of it was collected in silos. Being able to put that all together and being able to generate those insights and again, connecting that back to these patient journey maps, 100%. I think that's what AI, uh has done for us.

Speaker C: Yeah.

Speaker A: And because I have my product hat on now, right. I'm like, how do you make that process frictionless, easier for the people? Is it conversational? Like, just give me the feedback. Right. It's like, obviously there's hip all the things, right. Regulatory things. But I think this is the beauty of this moment.

Speaker B: It certainly is, yes.

Speaker A: Right.

Speaker B: I mean we always thought of m those things. We always wanted to do those things. But now we can do it.

Speaker A: Now it's possible. It is possible.

Speaker B: And you can do it fast.

Speaker A: That's right, yeah. Isn't that amazing?

Speaker B: It is.

Speaker A: What are you most excited about in the pharma space right now with AI?

Speaker B: You know, I think for me, clinical development was something that has always taken years to do. Uh, for me, if we can use AI to improve that, like getting the right patient on the right trial, getting the drugs to patients at the speed of their clinical need, AI can allow us to do it. That is something that I'm very excited about.

Speaker A: Okay. You come from a regulatory background, so I have to ask what's missing in regulations today when it comes to AI development and healthcare and pharma?

Speaker B: Um, we established that regulators are lagging behind globally. That's a given. I think it is going to be like this for the foreseeable future because evolution of um, innovation is happening so fast.

Speaker A: It is.

Speaker B: So I think what we need to have is some bottom line, some fundamentals, fundamental benchmarks, regulatory benchmarks. You might not be able to explain 100% of every single aspect of your model, although that's what you should be able to do. But then with models innovating so fast, at least being able to establish a few bottom line, um, assets, say for example, transparency of your model so your model is not black box. Being able to provide, uh, uh, really good data, really good data on what the intended use of this model is going to be and what are being able to look at your data and think about it in failure modes. Are you able to do adversarial testing of your models to an extent where you know where this model is going to break down and if and when it breaks down, what is the next step? Like, how are you going to test it? I think it's important for Regulators to establish those parameters and they are doing it across the globe. So I think that is something we need right now. Um, and I think for pharma companies, I think again, regulations are catching up. I think pharma companies in the past, the channels of communications have changed. Um, the way we engage with HCPs and other stakeholders has changed. Um, being able to. The regulations need to evolve based on that shifting trend as well.

Speaker A: So, yes, I have a separate sort of convergence question then I'm going to get into some rapid pub fire with you. We've had an amazing conversation. I knew it was going to be easy and we're just, it's like we just needed a glass of wine and this would have been perfect. Where does robotics or quantum fit into this? Right, I know we're now going ahead. Yes. And you may or may not have an answer. But you know, we know robotics plays a role when it comes to development of, you know, drugs. And quantum is just going to like constantly things. Is there any conversation of that or is that like, where do you stand on either one of those?

Speaker B: I think with robotics we already see it in manufacturing, we already see it in quality control. So I think robotics is already there. Uh, with quantum, um, at least in my personal experience, um, we look at it as our next step, as a future. But I don't think perhaps in R and D and in understanding, um, interactions, interaction of your drug with your target. I think quantum computing has become very helpful in that. But I think we have not had drugs that have ascribed their development completely to quantum computing. That has not happened yet. I think it is our next step. I think we are at a point where we are still understanding the full potential or trying to tease apart the potential of, um, quantum mechanics and quantum computing.

Speaker A: Okay.

Speaker B: Um, so that's where we are going right now.

Speaker A: All right.

Speaker B: But I think there's a lot of, I mean it's super interesting.

Speaker A: It is, I think that's, I just think there's this world of convergence happening.

Speaker B: Yes.

Speaker A: Altogether. Right. And it's um, it's a, it's an interesting place to be living in because we, we know the next 20 years are going to change so much and that feels so far away. But three years from now is going to change so much. So let's sit, let's start there in 2030. Right. We're in 26. So four years from now, what will be normal in healthcare? Because AI now, it sounds, it sounds shocking today, but it's actually happening.

Speaker B: I love this question because, you know, I want to go back to what we talked about, personalized medicine.

Speaker A: Yeah.

Speaker B: I think that has a really good chance of, um, being a fact in five years, in three years from now. Because I think now because of AI and uh, really how fast you could analyze data. Being able to take an individual's genetic data, genetic markers, combining with all the environmental markers, looking at the treatment history, looking at other types of real world data, et cetera, combining all of it, and being able to take from billions of such patients several billion data points, and you can analyze it in days and months. That's a possibility.

Speaker A: It is.

Speaker B: So that means for a given patient, for one patient, a, uh, customized dose for that particular patient is going to be possible in three to five years. There's another area that I'm very excited about, but I'm not sure if that's going to be a possibility in three years. It is about target biology. So for any drug, it has to have a biological target. Um, in the past, understanding how proteins fold was important. And that would take years for you to do in lab, because that is important when you're designing your drug, because your drug is expected to bind to that target.

Speaker A: Okay. Uh, you're educating me right now. So I love it. Keep going. I'm like, the protein folds. Okay. I'm like, let's go. Biology was never a good subject of mine. I'm just sharing. And chemistry. Any of those.

Speaker B: So anyway, I'm inorganic chemistry. Business.

Speaker A: I'm a business person. But go ahead.

Speaker B: So then it used to take years of lab research to do it. Now we can do it in days and perhaps in months. That's doable. That means you can now develop drugs, design drugs, do it so much faster. Um, and you don't have to test everything in a vet lab. In a preclinical setting, you could model protein folding, you could model the design of a drug. You could see how well they bind. And you could do that in days. That means you're easily chipping off ten years of time. Like you're saving yourself that much time.

Speaker A: Can you guys hear what that, just what that means?

Speaker D: It's.

Speaker B: Is it days and months, taking away

Speaker A: 10 years of time and development of.

Speaker B: Yes. And I do want to be cautious. I mean, it is, it's not. I'm, uh, just saying this would happen.

Speaker A: Yeah, of course. But, but that's our reality and the possibility. And that's why I want.

Speaker B: That's the art of possibility.

Speaker A: It's possible today.

Speaker B: Right. And it's. Think about it. Yes.

Speaker A: Because we have the tool sets, right? It's building. Some of the infrastructure needs to continue to be built, but some of the infrastructure is here. It's just putting the, I call them the AI Lego pieces and additional technologies together. Right? Um, so it's just. I love it, I love it so much. So I'm going to ask you a couple rapid fires, right? What do you think your most overrated AI trend in healthcare is right now?

Speaker B: I would say AI chatbots that are supposed to be symptom checkers. They have been here for many years. I think when they came in there was this, um, huge hype around how at some point they are going to take over, uh, your primary care physicians, et cetera, because they would be so educational and helpful to you. That hasn't happened yet. And, um, so I think that there

Speaker A: are other things that are happening, so that's okay. What about the most underrated AI trend in healthcare?

Speaker B: I think what AI understanding that, for example, being able to reduce your administrative burden, be it in clinics, uh, say for example, your regulatory submissions, et cetera, that's a lot of time each day you might be working on it for two, three hours. And AI has been doing it, AI has been giving you back, say for example, 20% of your time every single week. And I don't think that gets celebrated as much as it should. So that's been underrated.

Speaker A: It's the operational excellence or a part of the process management, AI can take over, which again, it's like people are here to build things and create things. I mean, we focus a lot of that with disruptive. Because they're like, why are you even doing this anymore? We can build systems for that.

Speaker B: Exactly.

Speaker A: So if, again, this is from our previous conversation, if AI gets a diagnosis wrong, people can die. How do we balance innovation and what's the trade off?

Speaker B: If I look at it as a all or nothing scenario, the trade off is a person can, as you said, lose their life. But I think I'm going to go back to what I said earlier and say the trade off is what is the current status quo and can AI improve that status quo? If it does, then I think we will have to take it.

Speaker A: Define a trustworthy AI system in one sentence. In one sentence. You got this.

Speaker B: You got this. A trustworthy AI system is one that spin, that knows what it can do. That knows what it can do. And when it is unsure about something, it knows how and where to escalate. And being able to provision it through a human expert, if an AI expert, if an AI Model or an agent AI system can do that. I would consider it trustworthy in 10

Speaker A: words or less maybe. What's the future of pharma and AI?

Speaker B: I would say precision medicine or the personalized medicine. Being able to develop medicine for each patient for their defined health condition. Looking at the person holistically, not just their medical condition but looking at where they live, what are uh, the different exposures, etc. Being able to do that, bringing the point of bringing medicines to you where you are. Where you are. Yeah.

Speaker A: I love it. If you had one innovation you could create using AI and forefarm, what would it be?

Speaker B: If I could, I would want patients to get their tracks at record speed. If I could go back and be able to improve the clinical development, the clinical trial process. So right patients get on the right trials and we are able to get those life saving drugs to the market faster. I think that's where I would want to innovate. I love it.

Speaker A: Listen, this is an amazing conversation. You know I again I knew it was going to be, it's filled with so much insights both, you know, I don't care if you're not in healthcare, it's just a uh, it's a conversation that's happening for every industry. Right. And these examples that we talk through are the realities of where opportunities lie, where some of the challenges, regulatory things lie. And regulation is really interesting right now because I feel like you have to go with your gut, you have to understand what responsible AI is and all the sort of core things from ethics and bias and traceability and all the things that you said. But in any system you should be focused on that. It's not just pharma. Right?

Speaker B: That is true and I think because healthcare is in my words at least the biggest use case for AI right now. I think what is happening is the onus to make sure that it's done responsibly is not falling on either startups or ah, established big players like say Meta and OpenAI et cetera. The onus in the past it used to be with the regulators. It was FDA's job to make sure that a safe and efficacious drug is coming to the market. Mhm. Now with these platforms because they are so fast developing so fast, it has become the job of these developers to make sure that a right platform, that the right platform that is safe and doesn't endanger patients is coming to the market. So I think that's a new mindset for developers as well. Uh, and I think the broader community is Catching up.

Speaker A: I love it.

Speaker B: So it's going to take a while, but I think we will get there.

Speaker A: I think we're going to get there.

Speaker B: Yes.

Speaker A: So with that, thank you so much for being on the podcast today. And for those of you listening, I have a couple PSAs. Listen like. We are growing and trying to reach as many people as we can. As season three grows, we're looking for the top 5% of business leaders in every industry that really understand what's happening in AI. To be guests at my studio here in Addison Morgan Legacy Sessions, uh, to have and engage in conversations like we did today, because they're gold. Uh, they're happening very proactively and how we engage and adding a lot of value to our audience. Um, we're excited that in a few months we're going to be launching something called AI CEO School to really help build fluency in another way. Not just around operational efficiencies, but what is really happening, where the opportunity lies. And AI. Uh, and listen, if you liked what you saw or liked what you heard, please subscribe. Like, share and comment, because this is that moment in time. I keep preaching, I keep saying it. My guests say it better than me. So appreciate you and we'll see you next week.

Speaker D: That's a wrap for this episode of,

Speaker A: um, the AI CEO Podcast. If you found today's conversation valuable, don't

Speaker D: forget to rate, comment and subscribe.

Speaker A: It helps me reach more leaders like you and continue bringing in top AI

Speaker D: first founders, industry experts, and visionary CEOs

Speaker A: to share their insights. AI is evolving fast, and the CEOs

Speaker D: who harness it today will lead tomorrow.

Speaker A: Stay ah, ahead of the curve, keep learning and keep building. Thanks for tuning in and see you next time. Sam.