AI's Prescription for Healthcare: Curing What Ails Us?

Julien Redelsperger: "And I'm super happy to welcome Shanil Ebrahim. He has a PhD in clinical epidemiology and works as a partner at Deloitte Canada. He leads the National Life Sciences and Healthcare Consulting business and works intentionally with public and private health sector organizations. So today we will talk about AI and healthcare. Thank you for joining me. How are you Shanil?"

Shanil Ebrahim: "I am very well, Julien. How are you doing?"

Julien Redelsperger: "Good, good. Thank you. Thank you for joining me. And so you have a very impressive background, I have to say, in healthcare. My first question for you would be, what drew you to focus on the impact of AI in this field? And also, why is it a good time to talk about AI and healthcare?"

Shanil Ebrahim: "Great question. So a little bit just about my background in healthcare. I started off as a bioethicist and a clinical epidemiologist, seems like many moons ago, and then eventually transitioned into consulting. But what has been at the heart of a lot of the things that I did was data. And what drew me into AI specifically was its potential to solve real everyday problems that we have in healthcare. You can imagine there's obviously tons of problems that we have within Canada as well, right? But when you look at the heart of things that a clinician does, imagine a clinician trying to predict how a disease like multiple sclerosis might progress in a patient. And traditionally, they'd rely on their experience because they have limited data and all they have is your subjective judgment or talking to their peers. But with AI, we can analyze patterns around thousands of patients to provide earlier, more accurate, insights as well. And ultimately, sometimes we talk about Al in terms of efficiencies, but I look at it as in terms of improving outcomes. And I worked on a lot of projects where Al was used to optimize hospital resources, like predicting patient surges and backlogs. But I feel that the impact that we can have on the patient experience, because now clinicians can spend more time on them, is just much more transformational. And as for, when you're asking, why now? I honestly do feel like the pandemic is what has pushed us to embrace digital tools. And then we've also seen this whole wave of gen AI really coming in in terms of how we approach data. So based on those things, I actually feel like we are at an intersection where we can actually create a huge, huge change and there's an ability for us to harness the momentum in healthcare with relation to AI."

Julien Redelsperger: "Okay, so that's interesting because ChagGVT was launched in November, 2022. And since then, it's like everyone is discovering AI and generative AI, but the reality is like it's been existing for a while, 50 years, 60 years. Did you see that change happening in 2022 when everyone was focusing on gen AI? And did you see new business case or use cases in the healthcare industry?"

Shanil Ebrahim: "Yeah, absolutely. So one of the things is you're totally right that in AI, even though we have made some monumental leaps in AI throughout the time, like deep learning is a great example, about a decade ago when we made a huge, huge breakthrough, the ability for us to use Al in mainstream was still relatively limited. And sure, you're seeing your apps getting optimized and all of those kinds of stuff. And in healthcare, which has been traditionally much more conservative, much more behind, the ability to end up leveraging that type of AI has also been quite limited. I think the difference between ChagGVT is it hit mainstream so quickly and everyone was starting to use it. And we started to see a lot of hospital users adopting AI immediately to ask very simple questions and being able to optimize a lot of their workflow. And so some of the clear use cases that we end up seeing are just ways to make lives easier, especially when it comes to documentation and documentation burden. So example, hospitals have tons of policies, tons of documents, tons of clinical information. And it's great to be able to insert all of those documents into a repository and using GenAI to be able to ask questions related to it, especially when you have a lot of standard operating procedures or different policies related to different clinical pathways. That's how we're seeing it used right now. I think we're still scratching the surface when it comes to using GenAI, because there's still some levels of privacy concern in terms of, are we using it in the most trusted way? And can we make decisions using Al without a little bit of context? So there's a little bit of harm that we have to actually think about. So that's what is taking a little bit longer within healthcare in terms of truly harnessing the impact. But I think the time will come and I feel like it's at least accelerated in the last two years."

Julien Redelsperger: "Okay, and so when you discuss with healthcare professionals, AI, doctors, nurses, surgeons, administrative people, what do they tell you? What do they think? And what's their mindset?"

Shanil Ebrahim: "Yeah, that's such a great question because healthcare professionals often have a unique perspective on AI. And when I talk to them, their responses are a little bit of a mix of curiosity, there's skepticism, and there's also excitement as you can imagine, right? Most of them see the potential, but just like I was talking about in the gen Al type of framing, a lot of them are like cautious in terms of how it fits their workflow. And a lot of physicians, for example, tell me they don't want to feel like Al is just another thing that they have to do 'cause they have a million different things that they have to do, right? And they're already stretched thin. So their mindset is, how does this make my job easier so that I can improve patient care? While unfortunately not losing any money 'cause they're also worried about that as well in terms of job security. So the good news is that when you show them Al in action, their mindset does shift. And I've seen clinicians light up when they realize, hey, AI can actually reduce my time that I spent on an electronic medical record and putting in all this information. But there's also a healthy skepticism, right? When I just talked to you about like trust, privacy bias in terms of AI making decisions that doesn't have context. So they look at it as an Al as a tool to support them and hopefully not replace them. And the mindset isn't just necessarily anti-AI, it is proof to me that this can make a real difference because my job is really busy and I don't want to end up just entertaining something that may incrementally improve things. I want to see something that drastically makes a change for me to actually completely shift the way I end up working. So those are the kinds of conversations that I'm having right now."

Julien Redelsperger: "And so what are the main problems AI can solve in the healthcare sector?"

Shanil Ebrahim: "Oh, Al has the potential to tackle some of the biggest problems, right? And I see like a laundry list of different things that we can tackle. But first you have to think about in the Canadian system, what are we in dire need of? Dire need of human health resources. And you might've heard of this news, like we have a shortage of labor and in healthcare specifically, there's a shortage of physicians, shortage of nurses, shortage of pharmacists, but also shortage of just backend individuals like technicians as well. So this has led to long wait times, overburdened systems and just shortages that have an impact across patient care. And that's where I see AI streamlining things. So we're using AI to now triage patients in terms of where they should go based on their condition. We're also automating like routine administrative tasks, documentation burden is a great example of it. We're also using AI to predict patient volumes and optimize staffing. So imagine knowing very well based on how many patients come in through the door, where are they likely gonna go within the hospital? And what is the amount of staff that you need at the hospital at the right point in time to be able to predict different surges? And that's huge, right? And so a lot of hospitals are kind of using these, or at least such as a testing these AI systems to be able to do that. But where I'm excited about, and this is where we're starting to have conversations we haven't seen at scale, we've seen it more in the US, is in predicting diagnosis and treatment. And of course, this is something we talk about quite a bit, but there's a little bit of skepticism in terms of ensuring there's accurate diagnosis and accurate treatment as well. And a great example of where this is truly applied is in radiology, where you can detect early signs of diseases like cancer much faster and more accurately as well. So they supplement radiologists in terms of making their jobs easy. So I think there's just a wave, Julien, where we're gonna start with operational efficiency and then it's gonna go to improving clinical workflow. And that's where I ended up basically seeing it. And then eventually where we start to get into, which we've seen in the consumer and the retail industry is like personalization. So Julien, you walk through the door, you have a very tailored experience from your treatment pathways to the type of doctors you end up getting, and it will just kind of revolutionize what it means to have a much different approach to healthcare. I think we need to be able to go through the waves to create the trust, create the traction before we actually get there."

Julien Redelsperger: "Yeah, you briefly mentioned radiologists. Why is such a prominent use case in healthcare and what's your perspective on that? How do you envision the role of the radiologists evolving in the future?"

Shanil Ebrahim: "Yeah, so I think you have to contextualize this within one of the biggest challenges that we're having in healthcare, especially in Canada. We have a huge shortage of diagnostic imaging. Everyone knows this because if you want an MRI, how long are you waiting? Very, very long to be able to get what is much more traditional in a lot of different countries. And because of the backlog, we need to rethink the way that we're getting access to diagnostic imaging, which means, yes, sure, getting the machines itself, but we also don't have the technicians as well as the radiologists to be able to scan it. So it's a huge backlog and a huge issue right now that we need to be able to address. And AI itself has shown such a high level of accuracy that can actually beat out human beings in terms of doing their work, which is why it's like this perfect mix. It's like massive demand, there's a supply issue, and it's actually much more accurate than human beings and it's not prone to errors as well. So I think that's why we're concentrating so much on radiology because it's not necessarily impacting direct patient care because there's still a human in the loop to be able to validate, yep, AI ended up producing exactly what I expect. I think there's a bit of a nervousness that is happening because radiologists are looking, they're very highly paid by the way, and they're looking at it as, hold on, will this replace me going forward? I end up thinking it will replace their work because AI can actually do their work a little bit better. Will it replace their job is a whole different story because radiologists needs to disrupt themselves to say, in the world of AI, what is my job going to look like where some of my work is gonna be done by the AI system? But will I be now overseeing a whole bunch of Al systems and be able to actually provide access to multiple patients? And that's the reform that we're still going through to kind of figure out where this is actually headed."

Julien Redelsperger: "From the patient's standpoint, what's going to change? Do you think they're gonna have better care, more time with people to talk to? Because I understand AI is going to change a lot of things in like behind the scene, in the administrative stuff and helping radiologists, et cetera. But what about the patients?"

Shanil Ebrahim: "Yeah, so, I mean, this is a great question, right? Because ultimately, if we end up leveraging AI appropriately, the idea is that it would end up taking out a lot of the burden and a lot of the administrative stuff that human beings right now do, or physicians right now do, so they can focus much more on patients. So ideal cases, yes, they would end up spending more time with physicians and getting a much more human experience with it. It will also improve their outcomes because of optimizing care and end up using data that can actually help them as well. Faster diagnosis, faster prognosis, all of that kind of stuff. Now, that isn't the ideal scenario. There's also the other world that we can actually look at as well to say, will patients actually end up interacting much more with Al systems as opposed to human beings in terms of diagnosis and getting optimal therapy? And I think this is the question that we're still trying to figure out because if we are now talking to a chatbot to be able to get that information, we are okay getting it for simple information. You're probably going to chat to BT right now to say, 'Hey, I have a headache and these kind of symptoms. What does it exactly mean?' If it's an appropriately validated health Al system, we're fine using it. But imagine you have a terminal illness like cancer and they're saying, 'Hey, we did the chemotherapy and a surgery and now you'll be dealing with this chatbot for any questions you may have.' For simple stuff, that's fine. But to get an experience that you feel like a human being is actually caring for me, that verdict is still out there. And that's something that we think we're not going to get there immediately because patients are going to push back. But when there's such a cost-conscious type of system that we have, are we going to end up heading there and what is going to be the pushback? And that's still something that we're waiting to see in terms of how the future actually unfolds. But there's a lot of kind of generational movements that are coming in as well, because people who are older may not like that. But what about younger individuals who typically interact with their phones right now anyways, and they don't want to end up having the human being to human being experience to the same extent. So I think we still need to see how this evolution is starting to look like. But in an ideal scenario, you would want the physicians spending more time on human-human interaction and all the administrative stuff to be actually tasked by the computers."

Julien Redelsperger: "Okay, that's interesting. You mentioned radiologists a few minutes ago. Any other use cases where AI could fit perfectly into the healthcare industry or the healthcare sectors?"

Shanil Ebrahim: "Probably one of the biggest ones will also be primary care so specifically GPs and family practitioners. Right now, I mean, it's very well known in Canada that we have such a shortage of family physicians. You know, one in five individuals in Canada do not have a GP right now. And this is a huge problem because we need to actually think of different ways to do it. A couple of ways that we have been actually testing out has been some are more certification and foreign workers type of related. So actually trying to get more human beings into our country so that they can be family physicians hasn't worked out as greatly, but we're still seeing the verdict on that. We're also thinking about new training channels like opening up med school in York University, where it's focused specifically on primary care and family physicians. And then the third one is now let's think about digital AI tools. And I think this is where the biggest potential is. So one digital mechanism that we're using is virtual care channels. Provinces like BC has done a great job in terms of expanding the service codes and using virtual care, which has been obviously accelerated by COVID. But the other thing that we're also starting to see is the use of Al systems. And the way that we're doing this, this goes back to my point around removing some of the work that is being done from a family physician and giving it to AI. So documentation burden is one of the classic examples. So every time you go to see a family physician, what are they doing? While they're talking to you, they're entering notes on their computer. That takes up more than half of the time of the patient interaction. So for every minute that a physician spends with a patient, two minutes are spent with a computer. So could you end up outsourcing those two minutes to an Al system, which actually alleviates two thirds of their time? So instead of actually finding three times more doctors, we can end up having the same amount of doctors, but two thirds of the work actually being provided to AI. And this is where you start to see systems like Amazon has their medical application, there's nuance, there's dialogue, all of that being able to record a lot of these information and transcribe and actually populate the electronic medical record within the computer. That could change the game. We haven't seen that at scale in Canada, but we've seen it in pockets in different hospitals. In the US, we're starting to see that pick up a little bit more. That's just a great example on how do you actually address shortage within primary care. And that I think has one of the biggest potential in terms of actually changing our healthcare system in terms of how we interact with it."

Julien Redelsperger: "Okay, that's a perfect transition because you have some experience in the US as well. Why do you think the US is ahead in adopting AI compared to Canada? And what's holding Canada back from a more aggressive adoption of AI in healthcare?"

Shanil Ebrahim: "Julien, that's such a timely question. Obviously, we've seen this whole Canada and the US politics, especially in the news right now as well, right? But I was actually on a podcast recently where we explored the differences between Canadian and US healthcare systems. So the topic is very, very fresh to me. The US is ahead in AI adoption for a few reasons. First, there is a massive private healthcare market just fueled by significant investments from tech giants like Google, Amazon, Microsoft, Apple, OpenAI. And if you look at the different examples on how they've actually leveraged and harnessed AI, Google's health AI has already demonstrated its ability to outperform radiologists in detecting breast cancer mammograms. And these are the type of advancements that are not just happening within the US, but they get funded and scaled quickly in the US by US governments as well. And I think that's a huge difference. Second is a regulatory environment that plays a big role. So the FDA has frameworks like software as a medical device guidelines, which creates clear pathways for AI tools to be approved. And this is where companies like Analytic, who has had AI platform, they gain a lot of traction and allows them to actually scale much faster. And this is only gonna get faster in the Trump administration, as he heavily favors a deregulated type of market as well, which is where Elon Musk is also playing a big role in terms of how do we actually deregulate it. And I mean, Trump just said it, what, last week or two weeks ago, that if there are tech giants who actually invest a billion dollars of jobs within US, there'll be faster approvals and less regulations. Those are the kind of incentives that we don't have in Canada. And I think that's actually gonna change the game just in terms of the dichotomy that we have between US and Canada. But in comparison, Canada has a single payer system, right? And why they have a single payer system is because they prioritize equity and accessibility. Everyone has equal access and equal type of allocation to different resources, which means new technologies have to prove a cost-effectiveness and impact on outcomes where the entire population could actually be adopted. And these unfortunately slow things down, but it also ensures that solutions are thoroughly vetted and thoroughly equitable as well. And in Canada, we're testing AI, like triaging in emergency rooms and doing pilot programs, but it does take a lot more to actually get it implemented. But I also do feel, Julien, that we talk a lot about systems in the political climate, but we know a lot of this is rooted in culture. In US, there is a drive, there's an entrepreneurial spirit. There's a drive for competition, there's a drive for innovation, there's a drive for IP, and that pushes AI adoption forward. In Canada, there is much more on fairness, ensuring access for all. And we have a much more of a social system that we're willing to pay taxes to get those type of structures. And that means often layers of extra scrutiny as well. And that's where you have to think about, there's private sector funding, faster regulations, competitive healthcare market, or do you have trust equity ensuring access? And with those different balances, you're gonna end up having a different sort of AI adoption curve across the different countries. And unfortunately, that's gonna end up holding Canada back a little bit in terms of being a leader in AI and healthcare."

Julien Redelsperger: "Okay, that's interesting. Yeah, thank you for sharing that. It reminds me of something happened during the COVID-19 pandemic a few years ago. I remember that labs developed and distributed a vaccine in record time. It was just such a breakthrough at the time. If another pandemic occurs, how do you see AI playing a role in speeding up the process of a diagnostic or developing a vaccine, for example? Do you think we would be more ready than we were before?"

Shanil Ebrahim: "I think we will be more ready than before, primarily because we have a lot of data to be able to predict patterns of where diseases are going to occur and where the pockets of spread are gonna end up happening. We always had those different models in place. Like I'm an epidemiologist, and this has been rooted in the last 100 to 150 years, where we have had models to be able to detect it. The issue was that it wasn't used at scale. So unless you have a lot of epidemiologists that were in your staff looking for public health outbreaks, like they do in Africa and so on, it's not gonna end up happening super fast. In COVID, it was such a, it hit the mass public so hard that you were starting to see data scientists actually dabble into this much more. So there was a lot of systems being able to actually predict what's happening. And it also helped a lot of organizations and new companies actually thrive in this type of market. So prediction is the first element to it, to say, 'Hey, there is an outbreak happening. This is actually much faster than we expected. How do we intervene?' So the intervention will happen faster. But what you're also talking about is, could we actually create treatments also faster as well? And truthfully speaking, the vaccines were developed so fast. It's actually still pretty crazy to think about it, knowing the dynamics and the mechanics of the science, because this should have taken years and years, and we were able to accelerate it. I can't imagine us being able to accelerate it faster, but you can imagine Al using a whole bunch of data to actually figure out what are the different antibodies and mRNA vaccines that can actually address that specific virus. And so you can still imagine that we're gonna continue optimizing it. And the third part is, how do you actually create adoption so that the right people get the vaccine at the right point in time? So the vulnerability can also be detected to say, 'These are the individuals who are most likely gonna benefit.' So those type of things are definitely readily available. I think we're well equipped. I think what's gonna end up holding us back is the political rhetoric. We have never seen viruses and outbreaks more politicized than ever before. And I think that's gonna be the bigger issue than AI systems or our ability to move fast in the world."

Julien Redelsperger: "Again, perfect transition, because my next question was about human behavior. And I was wondering, how could AI influence human behavior to be in a better health, to focus more on prevention than treatment? What are the main hurdles? Where do you see a way that AI could help us to be in a better physical and mental health in the future?"

Shanil Ebrahim: "Yeah, absolutely. So Al has the potential to shift healthcare from what we call a reactive or a treatment-based model to a proactive, a prevention-focused one. Right now, AI can analyze patient data to predict illnesses before symptoms even appear, right? And that's gonna enable us to intervene much earlier. We're already kind of doing that, right? I'm not sure if you have a Apple Watch, but a lot of people wear Apple Watches, which is powered by AI, and they can monitor heart rates, sleep patterns, and other metrics to flag potential health issues like atrial fibrillation or diabetes risk. And I'll give you a personal story on how Apple Watches have impacted my family's life as well. There was a few months ago when my wife was actually quite ill, and we just thought, 'Okay, well, it's another fever. 'We have two young boys. 'A lot of viruses actually go through the family.' And it was her Apple Watch that actually ended up sending out alarms, essentially, to say, 'Your heartbeat is way too high. 'Go to an emerge.' And it was around 145, 150, which is obviously very high, and no one checks her heart rate all the time. And so she thought, 'This is very unusual. 'It's not coming down.' So she ended up going in, and she had pneumonia. And she was admitted to the hospital and got it checked out immediately. And this was one where both of us are in healthcare, and we just ended up just kind of dismissing it as, 'Okay, it's just gonna be another cold or another virus.' And that actually helped her get earlier intervention quicker. So that's one example. Another example of, and this didn't actually happen, but I'm hoping it doesn't happen. My mom had a fall a couple of years ago, and it wasn't too bad of a fall. She ended up phoning us, and then we ended up getting to her, to a hospital immediately. But imagine if it was worse. And right now, Apple Watches are triggered that if a fall does happen, it can contact two numbers. It can contact your personal emergency number, which would be me in this case, and it would also call 911 so that it can actually dispatch a unit to be able to actually address what's actually happening. That is something that could prevent a lot of major injuries, deaths from actually happening as well. And for my mom, she ended up getting lucky that it wasn't too bad of a fall, but it could have been worse. And we would not have been there to actually care for her. So those are the kind of things that we end up seeing that could truly prevent a lot of illnesses happening. But I think there's a lot of hurdles, obviously, along the way. First right now is data fragmentation. So especially in Canada, health data is often siloed. It makes it harder to complete, you know, a picture of a patient's full health record. So that's a big one. Second is privacy concerns. We talked a little bit about it, but patients and regulators really need to trust their data that is used ethically. And I do think that we have taken way too cautious of an approach because actually limiting innovation quite a bit. And the last is the issue of healthcare funding models. Right now, most systems are still designed to pay for treating illnesses rather than preventing them. So imagine like if all of us were taking really great care of ourselves and preventing illness, what do hospitals do? They need to rethink the way that they're gonna be doing their business and what's their business model going forward. And that's actually limiting us in terms of how do we actually evolve as a system. So if we overcome these challenges, AI could help us predict and prevent many illnesses and ultimately improve healthcare outcomes and reduce the overall care. And I'm hoping that's where we end up going, but there's a few things that we need to actually do before we get there."

Julien Redelsperger: "Okay, okay. If you could focus to one AI-driven innovation for the next decade in healthcare, what would it be and why would you be particularly exciting about this?"

Shanil Ebrahim: "It would probably be AI-powered early disease detection because I really do feel that we still diagnose way too late. And I think if we diagnose things faster, people are able to get treatment much, much faster as well. And that's a potential to save millions of lives and just completely transform how we approach healthcare. So imagine a world like cancer or Alzheimer's or heart disease, where we're able to catch it much, much earlier than there today. You won't have stage four cancers anymore. You'll have stage one cancer, which is much easier to treat. It's less invasive, just far more effective. And we're already seeing glimpses of this in the future. Google's DeepMind, which I think also released something just last week, they are able to predict the structure of proteins with unprecedented accuracy and which paves the way for understanding and diagnosing complex diseases much, much faster. There's also companies like PathAI, who is using machine learning to improve the accuracy of cancer diagnosis, just using pathology slides as well. There's Zebra Medical Vision, which is using imaging for, or Al systems and imaging, I should say, for osteoporosis, liver disease, cardiac conditions as well. And this type of innovation isn't just about just giving people more time, but it's about saving lives, having a much better quality of life. And I actually do feel like we're gonna need way less treatment if we're able to reduce the severity and catch these diagnosis much faster. So if I was making my bets and I would wanna see one place where we push the AI innovation, it would be that."

Julien Redelsperger: "Okay. Do you think AΙ innovation is equally distributed across the country?"

Shanil Ebrahim: "The short answer of it is, it is not distributed equally. And that's part of the challenge. And it's unfortunate because, I was just mentioning that Canada's ethos is about equity. And unfortunately with Al systems or just AI innovation, it is major hubs like Toronto, Montreal, Vancouver, Edmonton, that has become a hotbed for AI innovation. Toronto has a vector Institute, Montreal has Miele, Vancouver has a lot of tech companies because they're on the West Coast with Seattle and California as well. And these attract talent investment infrastructure, which is unfortunately gonna leave smaller cities or rural areas lagging behind. And one big factor is access. Larger cities often have better funding opportunities. They have academic institutions like the UFTs or the McGill's of the world. They have industry players who are gonna be willing to collaborate. They have academic or non-academic institution. They have big hospitals like your UHNs of the world or the CHUM in Montreal, who's gonna end up actually collaborating with them to provide them the resources to adopt or scale Al technologies much more efficiently. So that's gonna end up creating an imbalance. And I think this imbalance is actually probably gonna increase before it decreases. And right now, what we're trying to actually introduce different initiatives across the nation is this kind of pan-Canadian AI strategy that's gonna aim to spread AI benefits much more evenly, which is gonna help businesses and public services and health care organization in less represented areas catch up. But I think this is how technology is always gonna end up playing a role. There's always some areas that is gonna push the pace. And then we're gonna introduce technologies like virtual care, like deploying AI systems in rural area who's gonna catch up. And then there's gonna be another technology that the new larger cities are going to end up pushing the pace again. So there's always gonna be a level of inequity, but it's also gonna end up taking digital AI to be able to actually close the gap. So this is a vicious circle that we're always gonna end up finding. And that's no different than we see in, not just in Canada, but in other parts as well in the US. Whereas a lot of the technology and a lot of the innovation happening, it's in the Bay Area, right? And we start to see them actually create these technologies, deploying it much faster. It's in a much more deregulated market before it starts to get into even other bigger cities like New York and Boston and all these other players as well. But it always takes a pocket of innovation happening in bigger cities before it starts to move forward. But I think this is where the government plays a huge role in terms of providing access to funding, providing the systems so that we're able to actually decrease the inequities that they have so that rural areas end up actually catching that as well. And, you know, Deloitte, you know, fortunately, and it's privileged to actually play a role in terms of thinking about how do we actually expand access? And that's a big part of how we work with different organizations to be able to do that."

Julien Redelsperger: "Okay, perfect. Thank you, Shanil. One last question for you. What do you like about your job?"

Shanil Ebrahim: "I think what I like the most about my job is the diversity of problems that I face on a daily basis. One of the most exciting things about consulting, people always talk about it being, you know, extremely fast paced, but the way I look at it is consulting firms are not cheap. So when organizations, large organizations work with consultants, they want to make sure that they get the value for paying those big dollars. And that means in the most complex problems that they have ongoing right now. So if we work with the biggest organizations on the most complex problems, we are working on the most complex, the most large scale, the most transformational problems across the industry. And that gives you a purview that you can't get in other industries as well. And that's something that I cherish a lot. I've learned so much, you know, since I've joined Deloitte now, you know, almost a decade ago, and the problems are still new. Like right now I am working on an extremely cool project where we are creating platforms that's going to end up providing access for a cancer therapy that is going to be a potential cure for a particular type of cancer. You know, that's something that I would not have thought of years ago when I ended up getting into consulting that this is going to be the role that I'll be playing. And I'm sure there's going to be these new type of problems that I'm going to constantly face. And so you're always at the forefront of some of the biggest challenges, and that's probably the most exciting problem, or sorry, exciting thing about my job right now."

Julien Redelsperger: "So there is always be a problem to solve."

Shanil Ebrahim: "There's always a problem to solve. And we're always going to be working with the most impactful organizations to help solve those problems. So we see ourselves as a huge partner in the ecosystem."

Julien Redelsperger: "Yeah, so at the end of each episode, the guests answer a question posed by the previous guest. After that, you'll have the opportunity to ask a question for the next guest. So here's your question, courtesy of Pete Pachal. He's the founder and CEO of The Media Copilot, a newsletter and podcast about how generative AI is changing the media world. We can listen to his question right now."

Pete Pachal: "So my question is, with A/B testing and changing, you know, how something is presented to a user and sort of gathering the data on that, what is the biggest barrier to having more sites, particularly media sites, adopting that idea? I think a certain amount of resistance because people think it might compromise integrity or what have you, but I feel like there should be more of this done in the media industry. But I also feel like it is something that quite isn't in the DNA of just sort of rapid testing. How do you achieve that in places that, you know, are more about journalism and media and what is the biggest barrier?"

Shanil Ebrahim: "Ultimately rapid testing in media sites and, you know, personalizing information involves a lot of data-driven approaches and tools like, you know, the person who was actually speaking, you know, mentioned a few things, right? Like A/B testing, multi-barrier testing, you know, personalization engines, user segmentations and things along those lines. And I think, you know, one of the biggest issues that we're actually having right now is that we are generating content at such a fast pace right now that it's, we're not putting enough testing to be able to actually, you know, test out that information. And I think one of the ways that we can actually solve this is actually through AI as well, because before a lot of the A/B testing, it had to be human coded to be able to actually test it, to actually see, you know, what the performance metrics are, whether it's, you know, click-through rates or engagements or conversions or anything along those lines. But now with AI, it can actually learn itself. So if you end up basically coding it in a certain way to say, I'm actually creating, you know, content or providing information on a site, it does the A/B testing in itself and it starts to refine the content based on the results itself so that it can actually improve rates, improve engagement, you know, improve conversions. Now, I think our ability to be able to do that is actually improving rapidly because now we're seeing Al agents being able to do things like that. I think the bigger question is going to be the ethical dilemmas we face, where when I'm a human in the loop and actually generating content, there's an ethical standard that I'm living by. But are you now giving the AI agent rules to say, I want to make sure that it doesn't produce harm? So for example, if I basically am providing you a, you know, like a therapy as a psychologist and I actually provide information on a media site and imagine Al system takes over to make sure that it has, you know, better rates, better engagement, better conversion, and actually provides you negative information, which has shown to actually create more click-through rates, more engagement, more conversions, it's actually taking into harmful behaviors. And that is a huge, huge issue that if you don't have a human in the loop, even if the A/B testing improves, quote unquote, better personalization, that personalization actually can actually introduce harm more than anything. And that's something that we need to be able to make sure of in terms of in the world of personalization, we need to make sure that we're actually providing benefits and not getting into a case that personalization actually creates significant harm. And that unfortunately is not governed appropriately right now, especially in terms of how fast we're moving with AI."

Julien Redelsperger: "Shanil, how do you keep track of everything's going on with AI? Because like, it seems that every day or every week, we have new features, new company, new startups, new announcements, could be overwhelming sometimes."

Shanil Ebrahim: "It is very overwhelming. It's actually one of the first times that I felt that I have been having an extremely hard time keeping up with a major waver, a major trend. I think in general, whenever we've seen other trends in the market, even previously with Al or blockchain or virtual care and diagnostic imaging, there's a lot of things that you're able to keep up with, what are going to be some of the biggest priorities and the biggest things that we need to keep track of. Now, there's so much noise, which is even exacerbated even by AI who's creating information about AI. And I think that's making it very difficult to keep up. So there's a few ways that I've done it. There's obviously just general research that kind of everyone does and just reading up on news, that's one, but that's not the perfect way to do it. The second way is through Deloitte, we have our AI Center of Excellences that actually curates this information and then provides us, these are the top 10 articles of the day or of the week. And these are some of the things that you need to consider if you're interested in these kinds of things as well, that really does help in terms of keeping abreast on, what are some of the biggest shifts that we're basically seeing and how do we need to move as an organization? And the third one, this is the one that I actually personally like the most is, I actually meet with a lot of thought leaders always. And I find the thought leaders are always ahead when it comes to thinking about AI on a daily basis. And through them, I'm able to understand not only what's the latest, but how do we actually apply it in the real world to actually create benefits out of it. And that's how I actually make sure that I'm using multitude of different market sensing approaches. And of course, there's other things like, reading different books and so on to keep abreast of the different trends. But those are some of the ways that I at least keep ahead of AI when I go forward."

Julien Redelsperger: "Perfect, thank you. So now what question would you like to pose for the next guest? Could be about anything personal, professional, healthcare related on that."

Shanil Ebrahim: "I would probably pose a probably a more philosophical one with the way that we have been talking about AI right now. And this is related to healthcare for sure. So the question is, if AI could extend human life indefinitely, should it? And what does it mean for our sense of purpose and the value of life itself if life goes on forever?"

Julien Redelsperger: "Wow, beautiful question, love it. Thank you so much, Shanil. It's been an absolute pleasure speaking with you today. Thank you for joining me."

Shanil Ebrahim: "Thank you, Julien, really appreciate it."

This transcription was generated by an artificial intelligence tool. It may not be 100% accurate and could contain errors and approximations.

‍