Data Analytics Chat

Choosing The Right Signal on Human Intuition or AI Insight

• Ben Parker • Episode 55

Share episode

Send us a text

Join us as we look into the journey of Sameer Sethi, Chief AI Officer at Hackensack Meridian Health, on the Data Analytics Chat podcast. Sameer shares his transition from financial services to healthcare, driven by his purpose. We see how AI is reshaping healthcare, the importance of human intuition vs. AI insights, and the critical role of human factor engineering. 

We discuss the pivotal moments in Samir's career and gain insights into the future of AI across various industries. 

00:00 Introduction to Data and AI Journey
00:29 Impact of COVID on Technology Adoption
01:06 Guest Introduction: Sameer Sethi
01:41 Sameer's Career Path in Healthcare
03:13 Motivation Behind the Career Shift
06:18 Defining Moments in Healthcare Technology
14:11 Challenges and Setbacks in Career
15:53 Importance of Self-Awareness and Mentorship
18:07 Balancing Technology and Healthcare
20:26 Advice for Technologists and Leaders
23:40 Introduction to Data Signals
25:08 The Role of AI in Job Evolution
28:18 Balancing Human Intuition and AI
29:57 Human Factor Engineering in AI
33:50 Ensuring Trustworthy Data Signals
44:09 The Future of AI in Organisations

Thank you for listening!

Sameer Sethi:

I started off just doing standard data and analytics, which is normalizing the data and adding business logic to it. Then eventually got into using that data in the form of signals in the form of automations that are built on top of this, and then now ai. So it's been a, it's been an exciting journey. When the whole world has seen AI is such a large opportunity, and as a result of that, organizations are vesting in providing folks like me and my team, the tools to deliver things that we could not think of delivering before. I think COVID changed a lot of things in the world, but it also. As us as a society embraced technology because we all had to be home for so long and relied on, cameras and computers to communicate with people and get work done. But I think the folks that are said, I'm gonna use AI and get better at what I do, I think they won't have a problem surviving at all. As a matter of fact, they will flourish.

Ben Parker:

Welcome to Data Analytics Chat podcast where we discuss the world of data, ai, and the careers shaping it. Today I'm excited to welcome Samir Cefi, chief AI Officer at Hackensack Meridian Health. In today's episode, we'll explore his exciting career journey and also discuss the topic of choosing the right signal on human intuition or AI insight. Samir, welcome to the podcast.

Sameer Sethi:

Hey, nice. Thank, excited to be here, Ben. Really appreciate it.

Ben Parker:

Brilliant. So obviously you've had an amazing career to date. I guess for the listeners, do you wanna just share your career, Johnny?

Sameer Sethi:

Yeah, absolutely. I tend to draw a line in my career back in the, towards the end of 2010 or so, 28 to. Two, 2009, 2010, where I was working in financial services. And I got into healthcare, i've been in healthcare since then. I started off doing EMR implementation, which is electronical medical record. This is the system that, that physicians and nurses use to record healthcare data about patients. And then I very quickly pivoted over to to looking at what we were doing with that data that was being generated. And since, and I've been here since, so I've been in, i've been working for providers. This is my th third health system. And and I took a bit of a break working in management consulting, which was really good because it taught me a little bit about healthcare strategy and how to build healthcare strategy versus just doing implementations. But for the last 17, 18 years, I've been focusing on figuring out how best to use use data. To to have organizations take better decisions. These could be patients could be, physicians could be healthcare operators. I started off just doing standard data and analytics, which is normalizing the data and adding business logic to it. Then eventually got into using that data in the form of signals in the form of automations that are built on top of this, and then now ai. So it's been a, it's been an exciting journey.

Ben Parker:

Amazing. And so I guess, what was the pivot, why did you move into healthcare? Obviously financial services, it's a good industry. Obviously a lot of investment in that field. What was the motivation to move?

Sameer Sethi:

Yeah I think that for me, there, there are two pinpoint. One is as I, I worked in financial services prior to coming to healthcare. And I married a very passionate, healthcare worker. My wife is an occupational therapist, and I think what she brought in me a bit of energy around being able to, or finding ways and means to do things beyond just a job as and and finding some meaning in that. So for me, I think working in healthcare is about either improving. The quality of care or reducing the cost of care, or making it more efficient, should I say. And I think that's been my mission. So I think that was one pinpoint, which is, looking at what I do and working hard at it and obviously, providing a living to the family, right? I think that was really important. But I think what triggered me was a conversation, I think with her around why I cannot do anything beyond just. Work, right? How can I convert that into work and make money obviously, and build a career, but then also help somebody? My mission is to help people with data and analytics. So I think that was I think that was one. And the other one as I mentioned earlier, was, I think when I started my career off in healthcare, it was about. Implementing workflows. And I thought that was fun. It was, I learned quite a bit about really seeing how, clinicians, nurses and physicians were using technology. But then the pivotal point was when I started to peek behind the scene, what I noticed was that there was, you could do a lot with that data that is being generated about a patient and not enough of that is happening. And that's what prompted me to get into the data and analytics. Game of ai. Of healthcare. So two pivoting points in summary. One is why and how I got into healthcare. The other is once I got into healthcare, how I shifted over to working on data and analytics.

Ben Parker:

Amazing. So the wife gave you the purpose. She found your purpose because

Sameer Sethi:

yeah. We're, she's one of those people that talks about patients and making people walk and, helping stroke patients and. And I hear those stories and I think a portion of me said at after a while that I bet I can do something here with what I know.

Ben Parker:

No, I think it's important to have that'cause. Work is obviously there's good days, there's bad days, there's medium days. And I think if you've got that purpose in the back of your, or what you want to achieve, it's gonna just amplify. Then tough days gonna make you push to achieve instead of just having a day off or put not pushing yourself. So I think, and a lot of people I think are. Not necessarily loss, but need to understand their purpose because I think a lot of people just turn out to work and not have not got that motivation behind them.

Sameer Sethi:

Yep. I agree. No, that's that, that helps every day.

Ben Parker:

Brilliant. So if you, obviously that's I guess, a good defining moment. Has there been any other like big defining moments in your career?

Sameer Sethi:

I think I feel like I, every time I find a new use case, an opportunity to tech enable somebody for me, that's a defining moment, right? Healthcare. Technology is still behind right. Quite a bit. And I think what excites me the most what and the reason I consider as a defining moment is because each of those opportunities that, you know either myself or one of my team members catches and says, Hey, technology can help, AI can help automation, can help with this, is is I think, a defining moment for me. And we. We, we I'd like to say that, we find at least one every two weeks or so, right? If not more frequent. So I think for me it's been I think those defining moments is what, keeps me excited and going every day, is. And there's no lacking, right? I think there's enough technology in healthcare but there could be more. And I think, especially now, I think those incidents of finding, how technology can help has only, I think the volume has just gotten up because I think technology overall has gone so much better in the last two years. Today we have generative ai, which we didn't have before. So I think I think that just opens so many doors for us. There's the concept of agentic AI right now that opens so many doors for us, right? To be able to deliver. I when I think about the things that, my team is, was able to solve compared to what they can solve today, I think it's amazing the progress that's been made. We can do a lot more, we can do things a lot faster. Using the help of, a lot of organizations that didn't even look at healthcare. So for me, I think, it's a combination of the defining moments are a combination of finding new opportunities, which we find every day. And then maybe, I think now that I think about it. What's happened in the last three years, two and a half years, right? When the whole world has seen AI is such a large opportunity, and as a result of that, organizations are vesting in providing folks like me and my team, the tools to deliver things that we could not think of delivering before.

Ben Parker:

Okay, brilliant. And you mentioned the sector is far behind and it is one of the, obviously it's health, it's associated to everyone. How can we, speed up

Sameer Sethi:

I

Ben Parker:

'cause I know, obviously I know the data is. It's really sensitive and it's, there's a lot of regulatory involvement in this field. How, how could we speed up?'cause then obviously if you can find more medicines and more impact, it's gonna make the world a better place.

Sameer Sethi:

Yeah. I think the way we can speed up is honestly, it's about the kind of attention that it has now, right? Which is healthcare. I think the, just le let's talk economics first, right? Look at the level of investment that's gone into healthcare. In, in the last five years compared to what it was prior to that. It's just it's amazing what's happened. So I think the first thing is for tech enabling. Healthcare, it requires investment. And that was lacking in the past. We are we're a nonprofit, we're a health system. The margins in healthcare aren't a whole lot compared to other organizations. And that's why, technology is not free, right? So it costs money. So I think I think bringing in the right kind of funding and the motivation to. To invest in, in healthcare technology is, I think what is happening already now, but it is more required. More of it is required, right? So I think that's, in my opinion, at a very large, macro level, I think I, I think that's what was needed and we've come a long way and more needs to be done. So I think that's one piece of this. The other other is I think, we are. Naturally, because we handle people and their data and their very personal data, healthcare data is very, personal in nature. I think what's what set us back is being able to make sure we, while we innovate, that we protect the data as well. And and that's hard to do both at the same time. And so I think I, I think that's been a bit of a setback for. At least tech enable or enablement or innovation. And again, I think, and things have come a long way, we have now ways to tokenize data and de-identify data a lot easily. So I think things have come a long way. But I think I think those are the reasons why, things are behind. It's really, investment and and then concerns around privacy.

Ben Parker:

Okay. Interesting. So has. The increase in investment, is that due to the pandemic?

Sameer Sethi:

I I think so. I think it's because of a lot of different reasons, but Pandemic had a play in it. I give this example to, to, to a lot of folks I know I have a I helped take care of a elderly relative and I remember going to her pre COVID. And asking her if she wanted to go for a, do a telehealth visit instead of visiting her doctor. And she would say, Nope. I wanna go see my doctor. I've been seeing the same doctor for so many years, and that's what I wanna do. Post COVID. When I'm asking her if she wants to go to the doctor, she tells me, can I do a televisit instead? So there's something to be said about in that story about the fact that people are now. Are comfortable with using technology as an enabler of healthcare. And I think there are ways to go with that. But that story tells you that that people are very comfortable with embedding technology as a part of how care is delivered. So I think that has a lot to do with it. So organizations have, tech companies and organizations have seen that. Momentum change, that the behavior change and does have invested in it because they can, make money from it. So I think that has led to a I think COVID changed a lot of things in the world, but it also. As us as a society embraced technology because we all had to be home for so long and relied on, cameras and computers to communicate with people and get work done. So I think I, I think COVID definitely has a lot to do with that. The other is, I think just in general, I think I think we are going through a tech revolution in my opinion, and it has impacted all industries and I think. You know what I'm feeling in healthcare, I think finance and other domains are feeling it as well. So I think that has accelerate, I think overall because processing of data and and digital capabilities have improved so, so much and so quickly that there's, there are things that to my, to, to, to my point earlier we can deliver that we were not able to deliver before. A good example is, 15 years ago when I was in this business, I we had to rely on a few AI models that were available in the market for purchase. And pretty much everything else we had to develop on our ourselves. But today we have models which are readily available, including large language models that didn't exist, right? There was no concept of large language models back then. So we have so much help from. From, for, from these enablers outside of our immediate, organization and our immediate vendor landscape. All that wasn't there. So I think I think it's a bit of COVID that has changed the way by which we consume technology, but also I think an interest from other organizations because they can deliver easily. A lot of foundationally we have, as a society, we have become very tech strong.

Ben Parker:

Yeah, no, and it's something, it's fascinating, isn't it? These big life events like COVID obviously has changed for many businesses. The working environment, obviously you've got a hybrid remote and that more as post before. And then, if you look at the AI boom, that's how many more businesses now are looking at investing in AI and things like that. These big events actually. I guess play like a domino effect, don't they? For people to start that we need to change.

Sameer Sethi:

Yeah, I agree.

Ben Parker:

Cool. So then obviously going back to your career, obviously you've done amazing, progressed to C level. Have you faced major setbacks during your career?

Sameer Sethi:

Yeah I think many, I, I, there's a saying out there, right? Every e every setback is a set up for a comeback, right? So setbacks are not a bad thing. I've, I have always seen setbacks as an opportunity to. To reflect on what what I could have done better. And it could be a position that I applied for. It could be a use case that we deployed that wasn't doing well, and the business came in and said, shut it down, because it's not really doing anything for me, and it's just a distraction. So I think I think lots and lots of setbacks. But I think each of those setbacks is, has, was a setup for a solid comeback because. At least the way I see this is you learn from that and you come back stronger. You built a capability that is better than it was. You learn from the mistakes. I guess maybe I sound different or, weird saying this, but I'm com completely comfortable with setbacks. It's a good thing. That's the only way we learn. That's the only way you really look back and think about, what you could have done better.

Ben Parker:

No, I think it's interesting. So did you get taught that? Or'cause obviously it's a mindset, isn't it, basically.'cause I completely agree, like any challenge in life is gonna give you an opportunity basically. If, whether you have a layoff, like it's always, you have that period of time where it's, yeah, it's not nice, it's a feeling. But when you look back, if you say, if you look back just a challenge you had 10 years ago, you'd think that was nothing. But at that point in life, it's like painful.

Sameer Sethi:

Yeah. Yeah. And to your question about whether I was taught it, I'm sure I was right. I'm a big believer of of mentorship I have folks that, that mentor me constantly and have, I've gone through so many of them through the course of my career. I think that's helped realize these things. Pe they're, these people are really smart and accomplished and they share when you have intimate conversations with them about their careers and their journey as such we're having here, they're very open to share what they did well and what they didn't. So I think those things have taught me, I think it was experiences or learning from other people have taught me things like this, which is, I think that setback is not a bad thing.

Ben Parker:

Yeah, so you must be, you must have strong sort of self-awareness skills.

Sameer Sethi:

I think so. I would like to think so. It depends on who you ask. But I try very hard to always reflect on what has happened and why it's happened, and be honest with at least, with everyone, but also with myself first, right? It is really important to be, as a part of your career or being a person, I think. You have to be honest with yourself. The day you're, you stop being honest with yourself is when you're doing yourself some injustice, right? And it's very easy to be honest with oneself, right? And that's what I try very hard to do. Reflect on what's happened, figure out what you can do better, right? Realize what the other, per other person's perspective. And I do this at work, not just at a, in a personal level, right? It's when I am trying to deliver technology I have to. It's very hard, but I have to see, try to see this in the view of a clinician whose life is different, right? I'm in front of a computer whereas these folks are in front of patients, right? So I think realizing that, you know how it's not just about tech, but it's about human factor engineering and that those concepts have to be brought into delivering tech. I think those, that's all a part of realizations and sometimes setbacks.

Ben Parker:

Yeah, no, I agree. I think you obviously you mentioned like review, reviewing yourself'cause you need E everyone wants to improve and there's always a better way to do things. It's being, again, I think it is like you having a blessing being really self-aware of where you are at and. Not lying to yourself, just being actually truthful where you're at. Alright, so what has given you a genuine edge?

Sameer Sethi:

I think what has helped me. Is the fact that I am open to other people's thoughts on, on, on a subject. I think that has benefited me a lot. I, I spend a considerable amount of time. At work and even outside of work, getting people's perspective on things, whether it be technology or otherwise. So I think I am blessed with with having a network where I can share ideas, bounce ideas, and that gives me the edge o over others. I think in a lot of people in that I see tend to. Sit in an I ivory tower that they built and feel comfortable being there. I'm one of those people that like to be uncomfortable at times and step out of my comfort zone and do different things and talk to different people and get their opinions and their feedback. And I think that starts to, give you an edge. I think that's really important. So I think that's one thing that's attributed to. Some of my success. The other thing is that I think understanding technology and the domain that you live in is the edge, right? I see a lot of people, know technology really well and others that don't know technology well and understand the domain really well. But I take a lot of pride in saying that I would like to be at that intersection of process, which is healthcare. Technology, right? And sitting in the middle takes, takes a lot of work, takes a lot of diligence, takes a lot of exercise, but sitting there and understanding both sides of this helps you deliver impact, and I've considered that to be edge. I do believe that I sit in the intersection of technology and healthcare, and that gives me an edge above people that don't have that position.

Ben Parker:

Yeah, and I, you touched on a key thing because obviously now tech. The tools now can do the heavy lifting. And data does feel like it is getting more domain focused. And you mentioned being an insect between tech and the domain. How, obviously as you progress to C level, how did you, or like even leadership, how did you find that challenge?'cause a lot of people have that. Move from a technical expert into leadership is a big challenge, a big step for people. How did you overcome that? So you mentioned you've had mentors, but is there any more advice for the listeners how you overcome them Challenges?

Sameer Sethi:

Yeah, I think number one I'll say is, and these are this. This advice is pointed towards. Technologist in general, right? Which is, and that is that I think you need to step away from your computer and walk the hallways, right? So I spend quite a bit of time, walking the hallways of hospitals. This is where you meet the nurses and you meet the physicians. You and you get a sense of what their life looks like. Because I think I'm very comfortable in the technology space and I can be in front of the computer and get myself up to speed on what's happening there. But I think I, I think stepping out of my eyes so I make it a point programmatically to do in healthcare what's called rounding. Rounding is a process where. Where you go and, go unit by unit. So you select a unit, you go in, you talk to them, look at how they're using the tools that have been delivered to them and then get feedback. Why this has not working. And you'll you'll see all the time this dashboard was built, but adeno, but no one's looking at it, and you figure out why, and then you come and improve on that. So I think getting outside of your. In front of your computer and experiencing the life of the people that you build these things for is, I think is I think really important. The other is networking, I think is important as well. I've been blessed with a with being able to, expand my network and, I think I have to I have to commend the leaders of Hackensack. They allow me to do this. Bob Garrett, who's our. Who's our CEO is very aggressive on making sure that we go to these networking events and shows and experience and learn from others. That has helped me quite a bit. So I think the advice that I have for people there is that, speak to your leadership, tell them what you're looking to do, and and give them examples of how you can just go beyond your organization and experience what others are doing. Network go to people's. Place of business where you can actually see things in action talk to people. I think so. I think networking helps quite a bit. So I think these are two things I would say. One is, one is, ex, experience. The actual business that you work for, get a sense of how they're feeling about technology that will give you an edge. And and the other is network. They're very smart people out, not just in the organization, but also outside the organization that are doing great things. And the only way you'll find out is either if you go to them and knock on their doors and talk to them. And again, share, share even what you're doing. I think those are two really powerful things.

Ben Parker:

Yeah. No, I love that. And because I think also educates you on their problems as I put those to not just seeing it from your point of view.

Sameer Sethi:

Yeah, I, look I was I even take, I even go as far as at least once a, at least once a year. Go to a conference that is not healthcare. So I was at a conference recently which was very general across all domains. And it was an AI conference and it was very powerful, right? That's where you got to meet people who are doing amazing things. May not all of them would apply to healthcare, but but just so I think what I'm trying to say is that step outside of just even healthcare networking events and focus on non-healthcare working events as well. And you learn a lot.

Ben Parker:

Brilliant. Okay, cool. So this, we'll move on to the data topic. And so we're looking at like choosing the right signal, humor and against ai. I guess before we, is. And.

Sameer Sethi:

Signals are, it's amazing if you look at the journey of data, right? Data comes in from multiple sources, right? And then, we data geeks, normalize that data and make that data usable. Then we add business logic on top of that data, and then signals come out of that data. The reason that is important is because signals is what matter, right? Otherwise data is not as useful as on its own. Data is actually not useful at all, right? You have to build insights, which is what a signal is, from that data. And the way you do that is, is by normalizing that data making it usable, and then adding business logic. Which is, which starts to tell you it's a good signal or a bad signal. And without that it's just numbers on an, if you, the way the way I explain it very simply is, you could get a Excel document and just put numbers on it and, I think interpreting those numbers are, is difficult. But when you start to add business logic, one thing, something multiplied by that equal to this, and then you add a, color formatting to which is red, green, or yellow. And if it's certain number, certain threshold, it turns into red, otherwise it's yellow and then it's green. Those are the signals. Is what makes this data most meaningful. So I think signals are the most important thing when it comes down to data. Without that, it's just numbers.

Ben Parker:

Yeah. And if, if we, obviously a lot more work is going down the AI route. Is this, how are we gonna deal with educating the team that AI is taking the jobs? How do you go, how do you deal with that?

Sameer Sethi:

Yeah, this, so that, that's an interesting conversation. I'll tell you how I'll start by saying this, right? I think I was just talking about Excel, right? I still remember I remember having this conversation when Excel came out, right? And people were saying, oh my God, Excel is here and there'll be no more accountants as a result of this. And last I checked, there are more accountants than. Than ever before, right? They're doing different things. They're not doing manual calculations, but they're focusing on some other things that we were completely overlooked. So I think I think as far as the jobs are concerned, that's I think there'll be some reconfiguration of roles for sure. But I'm not worried about AI causing job losses per se. That there'll be news about this and things will be twisted and turned in my opinion. But I do believe that AI is going to redefine how we work and what kind of work that we do. Today for the next five years, I believe AI is going to help. Organizations work at the top of their licenses, and that's what I am interested in. What we are, the AI that we are building at Hackensack and using is allowing clinicians to work at the top of their licenses. It is not being built or deployed to get rid of clinicians. That's not what we are doing. We're making sure that we can provide better care with.

Ben Parker:

No, I agree. Obviously there's gonna be changes, but like you said. Excel. Yeah. Has, it has helped accountants and, but it's still gonna be, there's still work. People still want to deal with humans, et cetera. And things will change evolution. Yeah, I just, obviously it's fascinating to you when you hear all these, like AI taking all the jobs on news, et cetera.

Sameer Sethi:

Yeah, it's gonna, it's gonna definitely reconfigure how we work and what kind of work we do, but that's okay. I'll tell you I spoke about, having mentors, right? And I still remember there was a mentor that once told me that. In technology the job that you have today did not exist five years ago, and the job that you have today will not exist five years from now. Right now the theme of that. That advice is that, we need to continue evolving, right? And I think that's what AI is gonna do. AI is going to force us to evolve as workers. There are things that we no longer will be required to do, and then we will move on to doing other things, right? And that's what happened with the accountants in Excel, right? Which is that the definition of what an accountant did changed. So I think in concept, the definition of. What we do is going to change as well as a result of ai. So no reason to be scared of it. I think the folks that are saying that they will not adopt AI at all and they don't think it's there to help them. They have to worry about things. But I think the folks that are said, I'm gonna use AI and get better at what I do, I think they won't have a problem surviving at all. As a matter of fact, they will flourish.

Ben Parker:

Yeah, exactly. Okay. So where should companies draw the line between relying on human intuition versus AI driven signals then?

Sameer Sethi:

Yeah. We, we think a think and talk about that a lot at Hackensack Media Health. I'm sure all health systems are doing that as well. And there is a line, right? Today when we think about AI enablement, it is about decision support. It is not about decision. So where we draw the line. Is where we need a human to take care of a patient. And it is that human is armed with decision support using ai. So I think that's where we, draw the line, right? Which is AI will not take a decision for a human but instead a human will take a decision for a human. But the ability for that human to take better decisions gets better. And that's when there's that line, isn't there? And then the way a human functions is different as well. And what I mean by that is that we're using AI to to remove things that are mundane or doesn't require a whole lot of thought, right? So the physician to patient interaction is something that is not managed by. You know that is not managed by ai, but things like documentation and summarization, things that AI does really well, we're letting AI do. So I, that's how we have thought about drawing that line of what AI should do and AI shouldn't do At Hackensack, it's a very much a decision support tool and not a decision tool.

Ben Parker:

Ah, that's interesting. So is that been a work in progress type of workflow for you?

Sameer Sethi:

Yeah, absolutely. As a part of our AI build and deployment human factor engineering is a very strong concept which is. Not only do we build a technology, but we look at the complete process that require, that has, as you can imagine, a patient and a and it has a nurse and a and a physician and other people that are involved, right? And we look at the complete process and then we make a decision, okay, this is something that can be managed by an agent or an AI and a robot or robotics process automation. And then these are the things that have to remain. So what we do as a result of. This design process is we redesign the whole thing, which is we look at we look at this as whole, right? To say and figure out where technology fits and doesn't fit. So that's been, that's that's something that was missing, I believe in the tech landscape in general, right? We, technology is believed in building technology without worrying as much about. As, as much about how it fits and how it gets used. I still remember, back in the day when, you know, when predictive health came into play, the celebration for success was accuracy. Which is we, I mean we, it was a bunch of data scientists that looked at the prediction and the percentage of a high accurate was, and that was the celebration, which was really good. And it's very much the case. But today we look at it differently. Today, we look at it and say, accuracy is important, but adoption is really important. And the way adoption happens is by looking at a complete process and making sure that signal that you mentioned Ben is delivered at the right time to the right person at the right place.

Ben Parker:

So do you think that's why a lot of firms struggle with AI adoption?

Sameer Sethi:

I I think so, yes. Because they haven't done human factor engineering, that, that's where a lot of AI is failing, in my opinion. I'll give you an example of this, which I think would hit home here even for somebody that, that may not understand healthcare in the audience. Think of so chat, GPT and the likes of chat, GPT came out, right? Which is a really powerful technology. And we gave and this society gave it to the to, to the masses and what it, but what it gave was what an empty box that you could type a question into, right? And you would type this question in and it would give you a response, what. But what we didn't do is that we didn't teach the users of this how to put together a prompt, right? So we didn't explain to the masses concepts of basic concepts of prompt engineering. As a result of that, people look at, people don't know how to ask the question. And so as a result of a lot of things happen, hallucinations and people aren't getting the right answer, and they've been disappointed with the response of it. The point I'm trying to make is that this technology was put out there. Without putting a whole lot of thought into how it'll be used or should I say how it should be used. That's it at a very big, and I think that's gotten better. People now are now understanding prompt engineering. We have AI to help with prompt engineering and all that is now waking up and eventually getting there. But the, my point with that example is that. There are various examples of technology that has been put out there without factoring in how it'll be used. And that is really important. Without that, people will use this technology the wrong way because they just don't know it. No one's told them how to do it. So it's I, in simple terms, it's about, buying appliances and or an or a gadget, and not reading the user manual. And when you don't use it, then you're not using the right features and you're not reaping the benefits of it.

Ben Parker:

Yeah, I agree. And it's, similar to I training for a marathon. If I don't follow a training plan,

Sameer Sethi:

Yep.

Ben Parker:

be hit or miss that I'm gonna get the time I want. But if you follow a plan, you've more likely got the proof that you're gonna achieve what is needed. So what, what makes a. Data signal trustworthy enough to act upon them.

Sameer Sethi:

So I think testing testing, we, again, depends on what kind of capability we are deploying, but I think UAT is very important, right? User acceptance testing it's one thing to to see things progress in paper. It's another thing to actually see it in action, right? So I think building a signal is very important. Building the right signal is really important, but how that interacts in the real world and in volume and at scale, how that works is a different ball game altogether. So I think for me, a good signal is a signal that, that is accurate as is accurate. Most of the time is something that scales really well and is built within the right workflow. That's what is a good signal any if you are lacking in either of these things, it's not a good signal or it's not a good enough signal, in my opinion. It has to be human factored. It has to be accurate. And it and it has to be something that, that people are. Able to pay attention to and find significant. A lot of this comes with training and a bunch of other things as well which change management and other things that have to be done. But without that it's not a good signal. To, to my point earlier, an accurate signal is not a good enough signal, right? A good signal is something that, that works for somebody. It's accurate, but it's also is delivered at the right time, right to the right person for the right purpose.

Ben Parker:

I suppose it's gonna be even more important for fish like new tools like ai.'cause it's moving away from the actual human element. It's more on the. So I guess that's gonna be quite key for businesses to, yeah, get close.

Sameer Sethi:

Yeah, I am actually really excited about it, by the way. I'm glad you're asking this question, Ben, because, when people ask me why am I excited about agent ai here's my spiel, right? My spiel is, for years I have delivered signals only to humans, right? And but now I get to deliver some signals. To non-human, which are agents. Now also, speaking about that line, right? It has to be, that line has to be maintained and not all signals have to be, should be actioned by an agent, but that's what excites me the most is when an agent is put in place or will be put in place, it's usually done. With process eng with a, almost like a forced thought of process engineering in mind, right? Because it's actually changing. You're actually saying a human doesn't need to do this step. Instead it's gonna be a machine or code that's gonna do it, and then a human comes in. So I think what I'm excited about is that it's going to force the mechanisms of thinking about deploying technology, but with a process focus. Without that, you just can't do agent ai, right? You just can't just throw an agent at anything and it's gonna work. You have to think about the whole process. So I am really excited about this because it, it's a forcing mechanism to think about process before deploying technology. Alright.

Ben Parker:

And yeah, also, it's going remove their mundane tasks for people. There's certain bits of your job that, yeah, obviously you wanna be good at your job, but there's certain bits that just, it's not interesting to people. So if you can get someone technology to do the. Bits that sort of don't need to do in your job your day gonna be so much more focused on the interesting, exciting parts.

Sameer Sethi:

Yeah, absolutely. And note also that, we were doing a lot of that already with robotics process automation, right? So this is, I, I, another way of describing agent AI is RPA on steroids, right? Which is it's actually now we have robotics process automation that can think and generate, right? So it's a robotics process automation that is fueled by a rule-based engine, and it is fueled by a large language model that can generate. Things and, statements. So it's beyond just I believe even mundane tasks. It definitely takes over the mundane tasks. But, we were doing that already, right? Using RP we should have been doing that using RPA at Hackensack. We've been doing it for three years. We almost have 300 use cases in production that, that are RP enabled. But now. This new capability is a smarter RPA. It's an RPA that can access various capabilities that I described, which is a large language model or a rule-based engine or some other capabilities, and we're really excited about it.

Ben Parker:

Yeah, no, it seems fascinating. Times ahead. So I guess on the importance side, how can organizations design systems that blend the human oversight with Ag AI decision making?

Sameer Sethi:

Yeah, I think I think you have to invest in, in just not technology, right? You have to invest in having people that understand technology and also understand process. If you look at my team today. We have technologists that understand, what an agent is from a perspective of tech, but then we have concepts of human factor engineering as well. Right? Which is which is really looking at the process and saying, okay, how will this be embedded into it? And all that has to come together. So I think the way organizations can benefit from this is having. Folks that you know, obviously that sit, in that, in, at that intersection of technology and process. And if you don't have that, then make sure those people are talking right? So don't build tech on your own. Make sure you have the business involved. Make sure this, it is their problem to solve and not a technologist problem to solve. So create the right level of accountability and ownership. Make sure that the solution that is built for them and not for you. And I'm speaking from experience. I'm speaking from a technologist mindset. I've seen too many example, not so great examples of situations where, you know, where an interpretation is made without the business around what problem needs to be solved and how it needs to be solved, involving the business. Solving a problem is what is required to deliver the best outcomes. And I think that can be done by hiring the right people that, again, sit in that inter at that intersection or, and or making sure the business is really involved in invested in solving that problem for you. Yep.

Ben Parker:

And I guess there, it's gonna be important to get. Hire people that understand the business problems, so having that domain knowledge.

Sameer Sethi:

Yeah, I, I, I was in a very similar conversation a few months ago, and, coming, coming off stage. I, what I realized was that it, it's amazing how much thought I have to now put into just not technology, but into human psychology, right? Into really thinking about how will this be adopted. So I think the role of tech has gone beyond, beyond code. I think the role of tech is, has, is now, is actually gone into how and how much it'll be used. And that's because so much of it's here, right? This, it's no longer a dashboard that we put out there, right? It's about signals and the signals are getting fired in different ways, right? So the role of technology leaders is not just thinking about technology. The role of technology leader now is actually thinking about process.

Ben Parker:

Okay, and say, obviously you gonna get. Businesses that do a thorough job, and it's always gonna be, in my opinion, people that want to try and cut corners. What, what are gonna be the big risks for, say, when businesses like blindly trust AI signals without like human inter interpretation?

Sameer Sethi:

Yeah, I it's,

Ben Parker:

because you, I think my opinion is people gonna try and cat catch up, get a competitive advantage. So it'd be, yeah. It's good to get your thoughts on this.

Sameer Sethi:

Yeah. It's a major risk and it's a major problem for us, right? We actually have a whole process by which we govern ai, which, which, and the way we do this is we look at every AI KPE that we either built or buy through a lens of a total of 13 domains. And the reason we do this is because there, this, the stuff is coming at us really fast, right? And I think their motivations beyond just, patient care that are involved, right? Because AI costs money, it is our job to make sure that we make sure that no one is harmed, right? Whether it be the user of AI or a consumer of ai. We at Hackensack, and I'm sure other organizations do this as well, look at AI from various different angles to make sure that it is safe. And so we tested technically, but we also tested non-technically. And by, by what I mean by that is we have folks in our ethics team looking at this AI capability. We have people in legal and risk, right? And compliance in clinical documentation. All of these looking at the ai. So we have a total of 13 domains that evaluate every AI that is either built or purchased, and then we help the organization take a right decision. Around whether to use that AI and not use it yet. The reason we put such a large investment into this is because there is, some that there are organizations that are releasing AI that is not a hundred percent mature, right? I wouldn't call it bad, but there is not a hundred percent mature, not ready for prime time, and it is our responsibility at consumers to make sure we test it to, to the best of our ability. To ensure that it'll not do any harm.

Ben Parker:

Yeah, and especially in your field, healthcare is such an important industry, isn't it? That you can't cut corners.

Sameer Sethi:

Yeah, we can, and look I wanna also say that I think that response be exists even in, not in healthcare as well, right? We have seen too many. Not so great stories about how AI was deployed and it hallucinated and gave the wrong advice, right? So I'm just stepping outside of healthcare and I'm asking people to really consider AI safety, AI responsibility. Before, before it is deployed. This is a very powerful technology and it need, it needs to be monitored, it needs to be regulated right? In different ways. And and that's the way to deploy responsible ai. Without this, people will get hurt and will get harmed.

Ben Parker:

Yeah. So looking ahead, how will AI change the way organizations act on data and insights?

Sameer Sethi:

Yeah. So I think I think that answer should always start by saying that AI is not magic, right? Agent is not magic. It is all backed by data. So the way I think the organizations. Are reacting to it should react more to is making sure that they invest the found, invest in the foundation. And the foundational, though the first, there are lots of other things when it comes down to, foundationally being ready for ai, but it has to start with data. So investments. So investments need to continue going into making sure that we have good data that the AI can rely on. So I think that is number one, right? Is which, is, which is the conversation of data and analytics in general. That cannot go away because AI has nothing else to function on. It does it's AI is fuel is data and data only. So I think that's one. The other is education. I think, educating people of what AI is and AI isn't is an interesting topic in my mind. And again, that has to attribute to, I think, things coming too fast, right? I think AI has come too fast at us and we haven't kept up with educating organizations on what AI is. My, my recommendation for organizations if they wanna adopt AI is keep that dialogue on, create committees, obviously that scale and and do things the right way, but. Having conversations with your leaders of what AI is and what isn't, and that changes and it's changing over very quickly. But I think that dialogue has to remain open, and very active. We as a, at Hack Sac Major and health, we, the, our AI and governance, AI and automation governance committee meets once a month. We have a working group that meets every two weeks. So such level of cadence is required if organizations want to go. And rely on, AI as an enabler of their success. And I don't think there's any ways around it. If you're not using ai, if you're not using automation, you will not be successful. You'll not be able to survive.

Ben Parker:

Yeah. Brilliant Samir. That's, it's been a pleasure having you on and provided some great advice. It's been a, yeah, obviously it's a fascinating time in the field and yeah, he said. We do need to make sure work is done properly.'Cause yeah, obviously we can't trust the technology. You need the human insight as well. So yeah, I wish you all the best with your work and thanks for coming on the podcast.

Sameer Sethi:

Yeah. Thank you, Ben. Thanks for the opportunity. This was great.