 Welcome back to SuperCloud 4, where we're digging into the power of generative AI and how it's affecting industry transformations and joining us is Scott Lycans. He's the global AI and innovation technology leader at PWC. Scott, good to see you. Thanks for coming in from Texas. Thank you for having me, excited to be here. So you got a cool title, set up your role. What do you do at PWC? Yeah, it's quite the title, right? When you think about an accounting firm. So I have the honor of leading a team that is focused on the future. So thinking about everything innovation. So obviously people are talking about generative AI. That's a big piece of what we do, but we really think about all technologies from what has been and is going on a blockchain, VR, AR, even quantum. So it's an amazing title. It's obviously a big role here at the firm being a global leader. So very excited and very honored to have this role. Yeah, so that's interesting. I mean, over the years, I've observed, you know, there is a spectrum of transformations, right? Sometimes a new leader will come in and change the culture or a company and pivots into a market. Those are kind of micro transformations, but the really interesting ones are those that affect entire industries driven by technology shifts that affect people in process and even the competitive landscape. And I think you agree, we're in the middle of one now. You know, Gen AI, of course, but you mentioned some others. 5G, IoT, virtual reality, metaverse. You said quantum crypto, new silicon architectures. There's a lot on the table right now. How are you thinking about this wave and the impact on industry? Yeah, I love the pace at which we're seeing this innovation. So I think one of the things we're seeing is just a faster acceptance. And whether that's because of COVID or just because of the phase we're in, you know, we've seen patterns of innovation over the years. I worked on the first browser in the 90s. I hate to date myself, but thinking about how the internet came about through the e-commerce web 1.0, web 2.0, now web 3.0 or maybe web 5.0, who knows? I think right now what generative AI has done is really gotten us broad acceptance that innovation and technology innovation specifically is happening and we need to accept it. And the pace is just unbelievable. You know, from the students that are inventing things every week to the big tech companies really investing in this at scale, this is amazing. And to be at a firm like PWC, we have such a broad spectrum of industries we support that's really what makes it exciting for me is how do we then apply that into the practical? So taking that innovation, looking two, three, five years out, but what can we do today? And I think gen AI has given us a really good vehicle to bring that to executives to say, the world's changing. You know, let's accept it, let's invest in it, let's take a chance on it, but let's make it very practical into business. So I love the pace. It's exhausting sometimes, but that's why I'm so excited about it. You know, because we're in the technology industry, we obviously focus on, you mentioned the first browser. So of course, you know, Netscape was all the rage. Most young people never even heard of Netscape. But the real winners of the internet were those companies that were able to take advantage of, yeah, obviously the technology industry, you know, exploded. And, you know, Google kind of came out of nowhere and social came out of nowhere, but you think about the economic impact of the industry transformation. So I'm interested in which industries that you see show perhaps the most impactful transformative potential right now. So when we think about, maybe I'll start with generative AI, I don't think there's an industry we're seeing that doesn't have some opportunity. So I can kind of go across those. I also want to think functionally, so we as a firm have been doing finance for 170 years, there's so much opportunity to think about finance differently. But some examples I think about is any regulated industry, thinking about how Jenny, I can speed up the process of working with regulators. So a pharma client, maybe being able to create a faster route to generating safety information about a new drug and getting it to market faster. So not only is it more efficient, it could be a revenue opportunity in the automotive industry, thinking about all the ways you interact with customers and being able to actually consume that and generate new product offerings or descriptions of your products at a faster pace. So again, efficiencies with revenue opportunities. I think about every industry we're working with, banking and insurance, massive amounts of information that they have to deal with either from customers or with their products, summarizing and going to market faster. To me, it's all about that speed and generative AI has shown us a way to do this differently. AI is not a new topic. I know people that have spent decades or their whole career in AI, I think these large language models really are showing us a powerful way to interact with information that we just didn't have before. So really across industries and then across functions within the business. The thing I talk to our clients a lot about is no matter if you're the CIO, the CFO, the CISO, the CHRO, there's value in generative models. There's value in generative AI. It's very different for those roles. But the same foundational model can help every executive think about their business differently. So that's where we're focused not only on the efficiency side, but on the change our business truly transform when you're thinking about finance transformation. Where do I wire this in everything I'm doing? Again, to be more efficient, have higher quality, but think about new opportunities for my business. That's, again, that's what we're working through as we speak in real time with our clients. What is the role of data in that transformation? I mean, it's a trite question, but obviously it's a huge role. But I guess my specific question is, do companies have their data act together where the stovepipe data fragmentation problem has been largely resolved and they can apply generative AI? Or does it not matter? Because in fact, as long as one of those functional organizations has their data act together, they can use Gen AI to improve their processes. What are you seeing in terms of the data estate? I love this question. I'm going to be a little contrarian here. I know that we've had a massive effort over probably a decade to get big data right, starting to organize it and create taxonomies around it. I think we're in an era where we're going to see a different interaction model with our data as humans, being able to talk to our data. You're starting to see this emerge now. To your point, these generative models or AI in general can look at both structured and unstructured data in a different way. So is there a world where we can actually put it all together and use new technology, maybe some aspects of Gen AI and other aspects of AI to really allow humans to interact with data no matter what it looks like? So I think we're in a really interesting place. I think that's where a lot of the innovation is happening. So having this AI actually walk us through the data as a human, I kind of feel like structuring data is for us, not for the machine. So I'm very excited of where this is going. I don't know that we've seen one answer, but to your point, there's different areas in our clients' businesses that they're seeing immediate impact of being able to interrogate and understand data. Like I said, in finance, we're able to look at financials, even though they're numbers, look at financials in a large scale and have the machine visualize it and tell us as humans, hey, I think this is what's going on in the data. Then as humans who are experts start to tweak and use that information or that visualization, it becomes much more powerful. So getting to that insight is going to be much faster in the future of being able to just talk to data. You know, when we look at the spending data, it definitely shows there's a lot of action going on, mostly experimentation, but there's definitely spending. You're saying AI spending, almost stealing from some of the other buckets because obviously the top line macro, IT's not just, you know, the CEOs aren't throwing money at IT, saying, no, go spend incrementally. It's more robbing from Peter to pay AI, Paul. And so my question is, well, at some point, I'm making an observation and a question, at some point the enterprise has got to show a return or the CFO is going to say, okay, enough experimentation. So is that return in your opinion going to come primarily initially from the back office, you know, productivity gains, labor costs, et cetera, or are there sort of front office impacts as well that you're seeing early on? So I think it's both. And I also think, you know, the technologists saw a lot of this coming. Transformers have been around, you know, called about five years, but the breakthrough, right? It came about, it kind of came in the middle of the budget year for a lot of our clients or even for us. So I think to your point, we had to find money to do this experimentation, but the reality is we could experiment very fast with these foundational models, with offerings both from the hyperscalers and from a lot of startups, we can start to experiment very quickly. And I think about my team working in one-week sprints, it's, you know, that's fast. So I think the budget year was a little bit of that, but now we have to find ROI. So the efficiencies, I think, in what we'd call the back office, so finding ways to do things faster and again, at higher quality is going to free up some more capital to invest in that true transformation. That's what we believe is the way forward. But rethinking entire processes takes, you know, a lot of guts to say, like, we're going to really change this entire process and rely on maybe machines to do parts of it for us. I don't want to think about just process automation of the current steps we have today. I want to blow it up. I really want to transform parts of my business. That will take more time. And to your point, we have to experiment, we have to experiment with ROI in mind. I almost stopped thinking about a proof of concept. I think about a generation zero, which gets me some value and a generation one, which gets me more value. But really focusing on the transformational aspects is really important, I think, and advising our clients to start with experimentation, but quickly move to returns on that and thinking about how you're going to weave that into core processes to get the value early and then sustain that value through revenue opportunities and true transformation. You mentioned process automation. Obviously, PWC has a lot of experience in process automation specifically, you know, around finance. You think about things like RPA. Do you think GenAI is a, it's probably both, but a headwind or a tailwind to some of the process automation software and the RPA momentum and sort of that. Now you see a lot of those companies pivoting to end-to-end automation. How are your clients thinking about that? Yeah, I think it'll be a creative in the sense of we truly want to get to intelligent processes and we want to rely on more technology. So robotic process automation was maybe that first step. I think we're already starting to see a lot of innovation in AI agents. So large language models on their own are not going to give us the automation. So we now have to think about that next wave and the next pattern. And that's where you're seeing things like AI agents that are going to help string together tasks and kind of break up process into smaller pieces. And that's where we'll see value. Even in our job roles to your point around finance, stop thinking about big roles and think about skills and which one of those skills can we automate and use the power of something like a generative model and which ones should be human-led, which we think is very important in this situation. We want the humans to lead through but then using the AI or the automation, the agents to actually accelerate some of the processes and again add quality along the way. So I think it's kind of that next wave. I think RPA was a little bit restricted because of the way it was interacting with a lot of legacy systems. This can interact right with the data. So maybe we can just get past those legacy connections and go right at what we're trying to do, which is automate a process in a better way, higher quality, faster with humans in the loop, enabling the humans to get to that creative judgment or that reasoning that we're so good at. We've been using sort of spoon feeding our audience with our power law of using a power law framework and basically just talking about a long tail. And as you go down that tail, you start to get much more model specificity, domain specificity, IP concerns and protection, being part of that, smaller models, et cetera. What are you seeing in terms of activity there? Are customers saying, hey, I want to do this work in my estate on-prem because I'm worried about IP leakage. I'm worried about privacy, security, et cetera. Maybe latency, you got the edge, which clearly is not going to be done in the public cloud, even though the public cloud vendors have a play there. Are you seeing companies actually realistically saying, we're going to do this in our own data centers or on-prem? So that's a tough question from, it's moving so fast. So maybe when you ask it, it matters. I think on-prem is still a little bit higher cost. You have to look at the amount of training in foundational models, trying to replicate that. But the breakthroughs we're seeing in open LLMs, the ability to have precision or fine-tuned domain specific LLMs that they really lower the cost to solve, that's where I'm focused. So I think it's going to be a combination play. There's power in what we're seeing in the cloud, in large foundational models, you cannot deny the training that's gone into them, the value that you can get out of them. But I also think there's a lot of innovation happening in the sense of how do we look at where there's some specific bespoke way we want to use an open source LLM. And we want to have some of our information that's very protected. But let me start back at the beginning. You have to have a secure way to do this. As an enterprise, you cannot allow your information to get out into open models, in my opinion. You have to build this in a secure and responsible way. So thinking about responsible AI frameworks around how we bring this in, whether it's in the cloud or on-prem. But I think the cost to solve is really important. So there's the training and then the inference and how broadly we're gonna use this. It's a very complex model. And we're trying to actually use some AI calculators to say, what does it truly cost? Because it's not free, we all know that. And then we should also think about the environment and what is the actual carbon impact of that? So all of those things matter and it's changing very rapidly. So in the war for GPU compute, there's gonna be a lot of different scenarios in an organization. And that's what we help our clients try and think through. There's not one model that solves everything. Although the foundational models give you a great start, you're gonna have to make some of those hard decisions. Haven't necessarily seen the on-prem momentum, but there's definitely some research and experimentation. But there's no doubt the foundational models are really providing immense value in the early days. And then I think as we mature through that, we get deeper into our processes. It'll be about embedding our data. I'm not, maybe fine-tuning, but mostly embedding our data, putting the controls and guardrails around it and then doing it responsibly throughout our organizations. And there's no question we're seeing momentum in the cloud because of the optionality and the innovation is there. Let's talk about scenarios, kind of art of war. If you're an incumbent firm, which a lot of your clients are, they get strong market positions and they're maybe either being attacked or they're worried about being attacked, how should they approach these opportunities and feel free to use industry examples or generic examples? Sure. We think of a few steps. First, get a secure environment. You have to be experimenting. There's no doubt. The amount of conversations we've had is amazing, but now it's time for action. So I think we've moved from the educational into the experimentation, but now I think we have to move into the enablement. So getting this in the hands of your workforce to see how they're going to use it is really important, but you got to start with a secure environment and whatever way you want to do that. There's lots of approaches. Then thinking about the patterns of things that Gen AI solves well. So before you even start use cases, think about the patterns you want to solve for. We're giving an example, summarization of documents. That's a huge pattern. We work obviously in finance, but across all of our clients, there's summarization of documents. Gen AI does that very well. Then start to break down the use cases by your business unit. So within research and development, can we start to interact with regulators? Can we start to interact with scientists? Can we start to interact with our own people in a different way and a more efficient, a quicker way through iterations? You think about marketing. It's amazing opportunities to start to generate multiple approaches at marketing. Maybe we get to more distribution channels than we could before because we were constrained by the amount of people we had. Thinking about the regulatory aspects of marketing in certain industries. You can kind of go through every business unit and start to come up with maybe the 10x use cases versus just the kind of 10% ones. And then we start to get into how do you scale this? So do you build a central AI, Gen AI factory? There's one large language model possibly that will satisfy the entire organization. Different than the days of machine learning where I can have different teams building what they need, I now have to think about how to control and govern this differently. And that way I can put in my responsible AI framework. So how do I scale this? How do I build a factory? And then how do I actually transform major processes? So starting to pick apart areas of my organization that I have these value plays and rethinking processes from the beginning and maybe scratching what we have and using Gen AI from the beginning to recreate a process. I know this is happening in software and technology right now. Can we really rethink the way we develop? I always joke, I don't actually know how to write Python but I can write it today. That's really empowering to me, right? I can actually write in any language and I love that. So software engineering is not gone but the flexibility and the capability, wow, that's really exciting to me. So we do have a way we think about this starting secure, responsible, thinking about the patterns and clusters of value, then getting into specific use cases, scaling it possibly in the AI factory or other approaches and then going in and transforming our businesses. And it would seem, I mean, all industries as you said are sort of ripe for disruption. But seeing healthcare is particularly an interesting use case where there's obviously, you know, a lot of tribal knowledge during COVID we certainly saw a lot of burnout, telehealth. And then as I say, you can pretty much pick any industry as you were saying before and there's a disruption scenario there. Is there any one that sort of really stands out to you that you're super excited about? I love a farmer life sciences, being able to get to market critical drugs and help to people that need it. And I think we're constrained it's such a regulated industry and it's a global industry. So thinking about how we use this to accelerate the time to market. I think that's just a no-brainer. I think, again, every industry has their opportunities but thinking about new ways to bring really complex things you mentioned, you know, the complexity the scientific documents and the understanding I can now read them, right? I can have it summarized in a way that I understand it. And that is really empowering I think for our organizations being able to really scale employees in a different way. But I think no doubt healthcare is one that I love but I can go through every industry and talk about opportunities. And my last question for you is we've been talking about gen AI a lot but up front we sort of both of us alluded to, you know different technologies as well. Are you seeing the combinatorial impacts, you know with gen AI complementing, you mentioned several I mentioned several other technologies, 5G. I threw in crypto because I love crypto, blockchain, et cetera. Are you seeing organizations actually begin to use leverage those combinatorial technologies or right now is it more sort of narrow application of gen AI? So I love this question. I think there's no doubt. We've talked about this convergence. We study something called the essential eight. So the emerging technologies we think are essential and you named most of them. But the convergence of them is where the true power comes in and I think the wave of gen AI has gotten the concept of innovation and technology being critical to every business on the board level conversation. So now we use that as the entry point to say, oh and by the way, you mentioned crypto, blockchain continues to evolve as a technology continues to break through in ways we can use it for moving money or value or payments around the world in a much better way. Can we actually decarbonize because blockchain is more efficient? You think about the metaverse, right? Everyone was last year this time, metaverse, metaverse, metaverse. Now today it's gen AI, but gen AI is gonna help us generate much more in the simulated world. You think about the multimodal concepts of generating 3D images, generative AI is gonna advance the metaverse in a way we haven't seen. If you haven't seen the Kodak avatars and things that are happening, it's amazing. We're even looking at what's next in neuromorphic computing and quantum being able to build better ways to compute, better for the environment, but much better in the sense of what we can produce and in compute. So there's so much innovation happening. I get back to that pace of change is unbelievable. I know we've had Moore's law for decades. I'm starting to see us break through that and I'm excited about that because it now starts to empower us with that convergence of technologies in a new way. So I love gen AI, I love the waiver on but what that gives us is that entry point to talk more broadly about what's happening in technology. Kyle, you're a great guest. I'd love to have you back. I really appreciate you participating in the program. Yeah, thanks for having me. It was fun. All right, keep it right there. We have an action-packed day of disruption, transformation, technology discussions, live and on demand from our Palo Alto studios, myself, John Furrier, Rob Streche, right back with Moore from SuperCloud 4.