GlobalEdgeTalk

AI in Regulated Industry Segments

• Alex Romanovich

Share episode

Send us a text

Dive into the complex relationship between artificial intelligence and heavily regulated industries with our expert panel representing government, healthcare, energy, insurance, and telecommunications sectors. This compelling roundtable explores how AI is fundamentally transforming traditional operational models while navigating the unique constraints of regulated environments.

Our distinguished guests reveal startling realities about our AI readiness. For instance, did you know it takes utilities up to six years to power a single AI data center? With extreme weather events already straining our energy infrastructure, experts warn we could face widespread grid failures by 2030 if we don't address AI's massive energy requirements through proactive legislation and strategic planning.

The conversation goes beyond infrastructure challenges to examine AI's role in augmenting rather than replacing human capabilities. Government implementations are already saving federal employees billions of hours by automating routine tasks, while healthcare applications are freeing clinicians from documentation burdens to focus on patient care. Yet these advancements raise critical questions about ethical implementation and maintaining human oversight.

The panel delves into the concept of "explainable AI" – ensuring algorithms can articulate their decision-making logic – and the importance of "human-on-the-loop" models where technology never operates without supervision. As one panelist provocatively asks: "Who holds the kill switch – humans or machines?"

Throughout the discussion emerges a central theme: successful AI implementation in regulated environments demands balancing technological advancement with human-centered design. By keeping customers at the core of development processes and implementing proper security frameworks, organizations can harness AI's transformative potential while preserving the human elements that ultimately matter most.

Join our conversation exploring the intersection of innovation, regulation, and humanity as we navigate the rapidly evolving landscape of artificial intelligence across society's most critical sectors.

Support the show

Alex Romanovich:

Hi, this is Alex Romanovich of Global Edge Markets and welcome to our virtual roundtable titled AI and Regulated Segments. Today we're sponsoring this event together with NetWeb Software. Ankit Shai is here with us. He's the head of growth and we have a very distinguished group of participants that I'm going to make some very quick introductions and let's start with Mark Buonforte, co-founder of Delta AI, former Ernst Young principal, also spent 12-plus years in the United States Armed Forces. We'll weigh in on the financial segment and government segments and artificial intelligence in those segments. Welcome, Mark. Thank you, Alex, excited to be here. It gives me great pleasure to also welcome Nandini Sankara, the Chief Marketing Officer at Suburban Propane. She also spent some time at Aetna. An executive there, sealed Air and Pitney Bowes and Nandini will weigh in on the energy and insurance segments. Welcome, Nandini.

Nandini Sankara:

Thank you, Alex, and look forward to joining everybody today.

Alex Romanovich:

Another great friend of mine, Stephanie Anderson, founder of SharedCMO, board member in a wonderful company dedicated to mental health, Solutioneyes, former senior vice president and chief marketing officer at Time Warner Cable, and executive with a number of companies like Charter Cablevision, Avaya will represent telecom and cable sector, as well as health and wellness sectors. Hi, Stephanie.

Stephanie Anderson:

Thank you for having me.

Alex Romanovich:

Isaac Chapa. Welcome. Isaac is a founder and the chief technology officer at Simple Healthcare. He also spent some time as an executive at Experian, CSID and a number of other organizations in healthcare and financial services sector. Will represent healthcare, finserv and weigh in on AI in those segments. Welcome Isaac.

Isaac Chapa:

Great to be here.

Alex Romanovich:

And last but not least, Ankit Shah, head of Growth and Business Development for NetWeb Software, one of the fastest growing IT services and integration companies in agentic AI with a number of solutions globally. Welcome, Ankit,

Ankit Shah:

Alex. A pleasure to be here.

Alex Romanovich:

I'm very excited and I suggest that we begin in this really wonderful and very interesting conversation at times edgy, so let's make it edgy, Mark, let's start with you. AI as an enabler, not a replacement. Let's talk about that and, of course, other members please welcome to weigh in on this. How do you see AI augmenting government roles without the workforce loss? I mean, we've seen recently a number of lost jobs in the government sector. But when we talk about compliance and documentation, when we talk about security and other very important topics, how do you see AI augmenting government role?

Mark Buonforte:

Yeah, thanks, Alex. So governments are making major investments in AI, and that's to reduce administrative burdens and free up their workforce for some of those higher impact tasks. There was a Deloitte study, and Deloitte estimates that AI could save federal employees up to a billion hours annually just by automating the repetitive manual tasks that take up the vast majority of these employees' times. For example, they're already using tools like Microsoft Copilot that have already saved tens of thousands of hours, and these tools are freeing the staff up to focus on more important things like policy and people right the high impact portions of their job, and this has been the focus of the AI tools that I've seen. In the data that I've read, ai is being used to augment human performance, but not actually replace the human, and that's what we're building at Delta. Ai, as well, is a tool for augmenting the workforce and automating manual tasks. Our goal is to make the work more effective, so we believe that we'll always need humans in the loop for judgment, ethics and trust.

Alex Romanovich:

Thank you, Mark. I'll switch around, and one of the probably most controversial and difficult topics to talk about right now is what's going on in the ecology and the environment energy. I do not have to, I mean, it's all over the news. I don't have to tell you about what's going on in Texas right now. What has been happening in major metropolitan areas, New York City, scorching heat, 100 plus degree heat, all kinds of outages, human life loss and so forth. Nandini, how can artificial intelligence scenario planning shift energy firms from reactive tactics to strategic resilience, to planning, to scaling, perhaps, and making sure that at least we cannot prevent the disasters, but at least we're ready for them?

Nandini Sankara:

Great question that's so timely. I mean, energy is probably one of the most, if not the most, polarizing topic, right, if you're going around, robin, we all have our opinions and what energy means to us. We all need it, we all use it and it's been so unfortunate with all the natural disasters. I mean what's happening in Texas right now I believe it's close to a $10 billion in damage and what just happened in the Carolinas. And you look at all the power outages in the state of California. And you know I've been in the energy industry now close to a decade and I'm still learning every single day. Then you throw in AI in the mix. Artificial intelligence needs data centers. We're all aware of that and you know I was at the BNEF summit. That's the Bloomberg Energy Summit. A couple months ago I read a really interesting stat Bloomberg Energy Summit. A couple months ago, I read a really interesting stat. The stat basically said it takes up to six years for a utility to power an AI center. Just think about that. Right, any given town and they pick them in remote locations, but in every state. You know I'm just talking about the United States even globally. You know parts of Asia, parts of Africa they're setting up these AI centers. If you will and Alex, you brought up ramp up time To me I don't think we're ready today, seeking Alpha just published something, I think, this morning, and I thought that was a very interesting statistic If we don't change our reliance on energy at the rate that you know, ai is growing data centers around the world by 2030, which, at this point, we're less than four and a half years away.

Nandini Sankara:

Right, we're up for a lot of problems. There'll be major energy grid failures, I mean with the 100 degree weather we're seeing right now, and also really brutal winters in parts of the world. Right, it's extreme weather. And then you're already having a power grid that's under a lot of pressure and you're pressurizing it even more. And so what does that mean for us as a country, as a world civilization economy? What can we do proactively, looking at alternative sources of energy. Right, I'm currently sitting in the propane industry and most places today, the way they power AI data centers without a power grid is natural gas power generators. But remember, there's remote parts of the country do not have natural gas pipelines. So one of the conversations we're heavily exploring is how LPG, as it's known globally, or in the United States, propane and butane. How can that be used, at least in the temporary phase? The temporary phase is not one or two months, we're talking up to six years, and even that may not help lower the overall reliance on the power grid, because you're putting all this no pun intended into powering this, right, but what happens to towns, cities around it?

Nandini Sankara:

So, in terms of being proactive, I think it needs legislations. As we're starting to build data centers, as we're actively moving, how do we manage that pace? How do we keep that pace At a federal level, state level, hyper-local level? We need legislation that protects the communities to make sure. Listen, we can't prevent natural disasters. They're going to happen, right, we're up and we're already in hurricane season. It started in June, ends in November. I lived through Hurricane Sandy. I see a couple of fellow New Yorkers here with me. He lived through Sandy. Hurricanes happen, nor'easters happen, storms happen, right, no matter how you look at it, it's preparing for this.

Nandini Sankara:

We need help from a government level. We need legislation that says, okay, you know what, as AI is ramping up, what else needs to be done? We need investments. We need investments in infrastructure, with energy. Right, we need to understand what are the different platforms we need to set up. What does that infrastructure look like? And it can't be a reactive thing, alex, to your point, because we're not going to be ready. We're going to fall way behind the eight ball, you know, not just here in the United States, around the world. So the main areas I look at is we need proactive legislation, we need proactive investments and we really need a scaled plan Like what's the short-term immediate impact, mid-term, long-term? And we continue planning for this because technology is just going to continue growing and evolving faster than I'm speaking right now. Right, so that would be my initial thoughts on this.

Alex Romanovich:

Yeah, thank you, Nandini. Very disturbing and challenging at the same time. It was interesting because the recent power outage in New York City. I was actually hoping for an internet operator In my case it was AT&T but I was hoping for other mobile operators, not only to give us some of the early warning system messages, but also was hoping for other mobile operators not only to give us some of the early warning system messages, but also maybe, with all the information that's being collected on us, maybe give me some kind of a personal messages. You know, alex, don't worry, it's going to come soon. Or you know we're in touch with National Grid and Con Edison and you know it's going to be okay, or whatever. Now, stephanie, you're a veteran of the telecom and cable industry. Do you think we'll ever see AI helping these giants? You know, improve our personalization, improve our customer service, if you will.

Stephanie Anderson:

Yes, thanks, Alex, absolutely. There's no question that we are well on the way and probably for the last I'm going to say six to eight years, a lot of fundamental work has been going on. That may or may not be considered AI as we know it today, but around machine learning and understanding behavior and trying to improve the customer experience and deliver the right information at the right time to the right person, and that's, you know, that's been the focus for many, many years. Now, again, a lot of that was pre some of the more popularized, popularized AI which you know, generative AI knowledge, and I think that's only going to help as well. But it's really. It's about implementation, it's about training, it's about making sure the customer experience is the best that it can be, and that's been the focus.

Alex Romanovich:

Yeah, it seems like AI is going to touch every part of our life, not only when it comes to disasters, but also when it comes to everyday type of activities. Health let's take a look at that, something that's near and dear to all of us. Creation and co-creation of new applications is going to be extremely important, Isaac, as somebody who is actually developing something for the healthcare end users and also partners. Is there a way that we'll ever see AI make its way into the clinicians and doctors offices, but also on patients' mobile devices? And how is this? You just need to contact us, so that's actually doable today.

Isaac Chapa:

One of the things I'll take a step back. You know, with healthcare there has been, unfortunately, a lack of focus on customer experience and clinician experience, right, I think one of the recurring themes I heard with some of the answers this morning were you know, there's really this short-sightedness of how can I improve what's going on today. And AI is the shiny object, but a lot of times people don't consider the other factors right. So healthcare has not considered patients and clinicians and I won't point any names. But a lot of the legacy systems are all about these behemoths that have been around for 30 years that just do not consider that experience right.

Isaac Chapa:

So it's very critical, you know, with AI as a tool, that we involve the patients, that we involve the clinicians in the design of those solutions, for various reasons.

Isaac Chapa:

Number one is really the biggest thing is they're the users of the system. You know it's not the big moneymakers, the CFOs, et cetera. They love to generate the revenue from it, but if the experience is ingrained with both patients and clinicians, there's more of an adaption to be able to use these products and they hopefully have better results, right? So we've already had some very successful results in the past year and a half automating a lot of tasks through AI, even through agentic AI as well, to help reduce the burdens that clinicians have by taking and writing their notes for them at a very high success rate, right? So you know, clinicians have to write notes for their encounters, etc. So if there's a way to get them to focus more on the care side of it as opposed to the document side of it, that will to the document side of it. That will help essentially everybody that's involved with the healthcare lifecycle.

Alex Romanovich:

Thanks, Isaac, Ankit on to you quickly as you're listening. Your company and yourself, you guys, have built probably every solution or every mini solution in the Gentic, AI and other solutions in every industry that's being mentioned here. But, going back to what Nandini was saying, we need data centers, we need scalability, we need security, we need to be ready, we need to be scalable. How do you foresee AI being scalable? How do you see companies being prepared and planning ahead, if you will, to be scalable?

Ankit Shah:

You're right, planning ahead, if you will. To be scalable, you're right. And as we talk about scale, you know the risk increases, right, model exposure, data leakage, you know attacks. So then, and especially in the case of implementing AI or AI-based solutions in regulated industries specifically, I think layered security is something that is very, very important, and when I say layered security, it starts from the AI SDLC itself. Right, I'm not calling SDLC, but from an engineering perspective, it is now AI SDLC.

Ankit Shah:

So, how secure, how robust you are having your AI SDLC, how robust your data security, your model security, right, model security is something that was not part of the SDLC earlier, but you're looking for biases, testing for any wrong data being there in the model, so that is very important. So, and then the CICD process how you are actually deploying it, how you plan to monitor your AI once it is out in the production. Is there any shift happening? Are there any anomalies? Right, how do you react to those anomalies? So I think there are these six or seven aspects that you really need to thoroughly map out, almost like a checklist, and when you do that, you know you can minimize a lot of the risk that comes in AI, even in highly regulated, high-stake environment.

Alex Romanovich:

Yes, indeed. And then, going back to you, you spent some time with Aetna in the insurance sector and you know I had an interesting experience recently with the insurance company as well dental insurance in that particular case. But what's interesting is that I guess I was part of some kind of a model. I was modeled because they assessed the risks, they integrated with my service provider or healthcare provider and then they either accepted or rejected a part of it, rejected part of it or whatever. But it seems like as consumers, as patients, we're all going to be scrutinized by every aspect of life, whether it's getting mobile service or getting health care service, or getting government service or being protected by the armed forces within the city limits or outside. Now, in this particular case, how do, as an example, insurance companies provide service, but they also provide fairness. How do we make sure that we're being treated fairly?

Nandini Sankara:

Interesting. We started off talking about energy and disaster. Right, that's directly connected to insurance, right. We all you know being a runner or a homeowner, and you know patient, we all need insurance, right, and I'm just talking about the United States. If someone's out there you ask that maybe different policies, if you will, or procedures, but in the US I'm a homeowner and you know what. I have homeowner's insurance, blood zone insurance because I live very close to the Hudson River, and we have health insurance, dental insurance, you know vision, et cetera, et cetera. Right, and just, we spoke about disasters a little earlier. So if you think about the insurance claims, and in many states and many locations, you can't even file insurance claims for most of the stuff that you were filing them on for disasters, right, especially if you're living way too close to the ocean. I mean, florida is a classic example of that. Right, and you spoke about health insurance, alex. I mean, I was just doing some research earlier and I think reinsurance costs, like mid-year, is going up up to 12%. That's pretty significant if you think about it.

Nandini Sankara:

Right, and just, with AI and the insurance industry, I think, even as a consumer, as an end user, ai has made our life to me a little simpler. I mean, who here doesn't use chat, gpt or some form of AI, if you will, to get your information? I know I've used it about 10 times today already. It's easier to get insurance and I think this is going to keep, and I haven't been in the insurance industry for a little bit. I was there when, you know, the pre-Medicare stuff was just really heating up and I know insurance, like energy, is always a hot potato item. Right, there's Medicaid, there's Medicare, pre-medicare all different types of insurance if you will, and, trust me, you need a PhD just to get through them. I spent two and a half years I'm trying to kind of figure my way out to my basic health insurance, but I think AI can simplify a lot of it. I mean, now that you brought up the model right, so there's a great opportunity to say, okay, this is the pilot program, these are your benefits Also as end users, and it could be a B2B or a DTC, a consumer like all of us, or business owners I know we've got some business owners on this call as well. What insurance does, what AI does? It could help simplify the process and I think insurance industries are really embracing that. I mean, I know, stephanie, you're in the education field, right, in education, you're just going ahead. You're talking about insurance across the board.

Nandini Sankara:

Ai helps simplify. It helps, you know, make it more accessible, make us understand better, and that's something that's lacking today, right, how many of us, how many hours of each one of us, spent on the phone, right, 1-800-figure-out-your-insurance-company? Right Went online and I think, from an insurance company standpoint, they've got to up their game as well. Because I know we spoke about legislation and regulation, a little bit about energy. Insurance is another industry that's heavily regulated, right, especially here in the United States. How can AI help some of that and, state by state, that could change, make that information?

Nandini Sankara:

I think, alex, you hit on something earlier with communication, and I think what AI is going to do, the number one thing that comes to mind right away for me, is simplification through communication. It makes it easier access to information, which today is a problem. It could be even pre-life event, post-disasters. None of us are perfectly prepared for the next hurricane. Next fill in the blanks right, it's fire if you will, but what if you have all this other stress? Whatever just happened in California a few months ago? People lost their homes, their livelihood. That insurance claim probably was the toughest thing, but it's a must do to get something some semblance of their life back To me.

Stephanie Anderson:

Ai can help with that from an insurance company standpoint and from an end user standpoint, overcorrecting one way or another in these insurance models or, quite frankly, any model but health insurance or natural disaster type insurance. I mean, I think that's going to be a very touchy area.

Isaac Chapa:

Yeah, and just to add on that, I don't know that it's biased, but I think a lot of times certain businesses may not have an ethical stance on how they use AI. We've seen issues in the past year that are raised on the healthcare insurance side, and if you look at the data on rejections, a lot of companies were finding AI vendors with the premise of help me deny more right. They wanted to increase denials. That shouldn't be the focus of AI. I love hearing the simplification and the abilities it can do, but if a company's sole focus is help me reject so I can make more money, that's not an ethical use of AI. That needs to be addressed.

Isaac Chapa:

You know, if it's regulation, great. I don't have too much faith that government regulation can help with some of those things. But you know, maybe it's things like hey, there's a certain standard of pass rates and if you're rejecting at a very high failure rate, maybe you should be looked at in your use of AI, right? So I think that's one of the big things to look at is the ethical use of AI as well. With these solutions, it simplifies, but it should have the intent of making it easier for everybody, not rejecting and increasing profits.

Nandini Sankara:

I think, Isaac, that's a really important point, because I think ethics and AI should go hand in hand, regardless of industry. I mean, you brought up a great example with really humanizing it. I don't think that EQ level of AI because at the end of the day it's machine learning right Are we taking the EQ away? That's applicable to we're representing a lot of different industries here. There's that emotional quotient that we should not, and ethics is directly correlated to that. We should not take that away. So I think you're spot on.

Ankit Shah:

And that's where I guess this whole concept of explainable AI is really coming to play, where you have to put explainability into the model so you know how it is kind of making decisions or deriving the logic it is, you know. So that's a big focus, even in compliance when you're implementing AI-driven solutions, of adding explainability to the model.

Alex Romanovich:

You know it's funny because, when you think about it, ai is this latest innovation that's been brewing out there for quite some time.

Alex Romanovich:

But it seems like you know at least I feel that way it seems like we're all now being a part of their focus.

Alex Romanovich:

Ai is there to either enhance what we're doing or replace us, or make our life better, or provide more services, or take away some of the services, and so forth and so on. So how do we make sure, as leaders in your respective segments, how do we make sure, as leaders in your respective segments, how do we make sure that we also build some human empathy and focus on human in this entire race for new innovation, for new tech, for a new way of doing things and sometimes, you know, doing some very inhuman activity, if you will, whether it's replacing you or taking services away from you, or automating some of the services around you, or something like that, how do we build, whether you know, if you're government, how do we build that human not just ethics, but empathy, excuse me into it? Or, if you're in a customer service environment, in telco, in telecom, in healthcare and financial services, how do we make sure that we still remain human?

Stephanie Anderson:

It's a you know, definitely a customer first.

Stephanie Anderson:

You know kind of view.

Stephanie Anderson:

You have to keep the customer at the core of everything that you're doing and that experience and you know you can still save money, you can still make money and keep the customer at the center of your business, whether it's you know telecom or cable or you know insurance, whatever.

Stephanie Anderson:

I think that we get caught up a lot of times with the fascination around the data or the information and we head in a different direction. But if you keep it to the point of having a human factor involved, the human is the customer and at this point I think if we kind of keep that at the center of everything and the other thing is too, as you're working on AI throughout your business, to also balance that with technical people being involved, and of course you need your data scientists and all that but you also need business leaders involved as closely right in lockstep, to define the strategy and set the pace and the tone and what the outcome is going to be, so that everyone is equally invested and working toward the same strategic goal, and that should be a better customer experience, so they'll keep coming back.

Isaac Chapa:

Yeah, I think building trust is the big key, right, and I think trust comes with time and data, right. So you know, to kind of get out of the segment of what we're in being in Austin, we're kind of an epicenter for the self-driving car wars going on, right. I've been able to see one company over the last let's say, 18 months slowly grow their footprint of automation around the areas. They made stupid mistakes early on, but it looks like they've learned from them and now you don't know that. There's, you know, 50 to 150 cars in Austin that are driving themselves and you see them all the time and now they've become part of the culture. You have another company who is how fast can we launch? Let's get it out ASAP already having accidents within the first couple of days.

Isaac Chapa:

To me, that's not building trust for me, right, I don't want to be near a car that I don't trust, right, I might trust another one. So, with me and I think, people in general, you want to get whatever AI you're using to build trust with you, right, using to build trust with you, right. So when we work with clinicians and patients, it's not a matter of here's a solution, go use it tomorrow, for all you know, 1,000 clinicians in your practice. We need to build the trust. We have to show them its capability. We refine it, we work with them to say this is what it can do. We get that trust from them showing its capability and then roll it out more right. So it's not a how fast can we get to market it's. Let's build that trust so people can use our solutions across all of our industries.

Mark Buonforte:

Yeah, and I think one of the ways to build that trust when it comes to using AIs is the human on the loop model, which is where the AI is just not out there doing its thing and making its own decisions, but there is a human that's providing input and refinement in real time, and then for organizations to broadcast that like hey, the AI is not just out here making its own decisions autonomously. There are people involved in real time in monitoring the decisions that are made to ensure that they're in line with whatever our procedures are, whatever our best practices are, whatever we want them to be in line with. So the model is not just doing its own thing.

Ankit Shah:

On a lighter note, friends, I was talking to a few people and you know how, when we go to websites right now, the CAPTCHA makes us make sure that it's not a bot and with so many bots, it's not very far when it's going to ask are you human? If you're human, you're not allowed, you know. So I think I agree to Mark. I think the human in the loop you know where we actually intervene, where we put the kill switches is important. You know, whenever we are doing AI implementation and we hear of the humans connecting to the LLNs right, they're connecting social bonding. You know it's not obviously very healthy, but that's something that is part and parcel of all technologies. You have the good users and you have the negative users, so it's a balance I think everyone has to kind of follow.

Nandini Sankara:

That's the ethics behind it too. I think, Isaac, you brought up ethics a little earlier. What does that algorithm look like, right as AI starts dominating things little by little? It's been around a while, if you think about it. I think, Alex, you kicked us off with that, but it's in the news now. We're here today. We're talking about this representing a plethora of industries. Just because it's a real thing, it's a blessing, but it's also a curse. So that's where the human versus machine. How are we drawing that line? And, to your point, Ankit, at what point are we? Who's the kill switch? The human or the machine? How does that look? So that'll be a very interesting conversation. As you know, this continues building and evolving.

Alex Romanovich:

Well, I would like to thank everybody for this amazing discussion. As leaders in your industry segments and people with a lot of experience, I think the humanity is depending on you and on the decisions that you're going to impact. With that, let's open the floor for some of the comments and some of the questions from the audience, but I would love to thank all of you for participation and, I think, the future, with you at the helm in your segments, the future looks very bright. I am not as concerned as I was before this conversation, so thank you very much, thank you.