 So welcome back to SuperCloud for our fourth episode focusing on this topic of generative AI, the focus of today's live in studio event. We couldn't have this next guest in studio with us today, so we're bringing them in remote. Vince Kellan, the CIO of UC San Diego, University of California. Vince, thanks for coming on theCUBE here for our generative AI SuperCloud fourth episode. Great to be here. Really appreciate it. Want to get into the practitioner side of it. You're out there, you're seeing a lot of things. CIO, you know about data, how data is structured. Generative AI is all the hype, all the rage. The reality is quickly matching up to the hype and use cases and how people are thinking about their IT environment, how they're organizing their data, what's being data, warehouse is being disrupted with data in the cloud. You're starting to see things like Parquet and Iceberg open formats, changing how data is going to be managed. You're starting to see data models that are starting to match with cloud scale. This next generation which kind of points us into this generative AI which just dropped out of the scene and mainstream over the past year. What's your perspective? Because you know whether you look at a learning system or an educational system or an enterprise, the game is still the same. You got content. You got to put stuff out there and engage with the customers. How do you see the gen AI rolling out right now? Because you've written some great papers on this and essays about knowledge graphs and about horizontal vertical learning, right in line with generative AI. What's your take of what's going on in generative AI today? Yeah, I think the thing that guides my thinking is generative AI makes knowledge generally more available to people. And it actually levels up the more novice learner, the learner who's newer and brings them up faster. That's really the big impact for generative AI. So I just kind of follow that thread across any dimension of a business or in our case universities and students. So it's a big deal, no question. Some of the early research out of the occupational side is showing that it's a very big deal in terms of improving both quality and throughput of answers to questions and support scenarios. It's certainly having an impact on, as you saw in Hollywood and talking to some folks I know that are in the script writing business. Obviously they're starting to use generative AI to help in all of that process. So it makes knowledge generally available to people. Follow that thought anywhere you go. I want to give you thoughts. I know you've seen any waves of innovation from a technical perspective as well as business and certainly education now and many years ago. When the web came on, we saw a similar dynamic. You had the internet was a transit, you had packets moving around all over the place. The worldwide web comes out and then you get websites. And that was a transition where everyone was disrupted. Oh, the web, should we put stuff on there or not? Is it browsable? We put transactions. We know what happened next. It all became adopted and happened. Education is online, businesses went online. Everybody went online, the online population grew. Similar thing now with AI, you have AI apps coming. And there's an infrastructure element too. So you had infrastructure, the web and internet. You have packets around that enabled the web and websites, which is the application. Here with AI, you have the same thing. This is changing the game on how users are going to get content and expectations are changing. Everything's changing with AI. This is happening now, the silicon's there. The cloud scale is there. The data architectures are there. Do you see a similar path where the entire world will move to be AI enabled at some level at an application, whether it's an AI wrapper app or what's called the AI wrapper, chat GPT kind of wrapped into it, or a cloud AI native app or something new. What's your take on this? I believe that many, maybe even most of what we know as our user interfaces today are going to change to more of a conversational style, natural language style of interfaces. So the metaphor I use is Iron Man asking Jarvis a question. So it's causing us to lean into what we call conversational analytics rather than fire up a business intelligence tool and start to finger click away. You say, hey, what were my sales last quarter? Or what were my student counts this last term? And suddenly you get your answer back. Absolutely, this is going to appear everywhere. And anybody who's making money as a software developer on their software interfaces is extremely concerned about and leaning into that conversational AI enabled interface. That said, costs have to come down further. Yeah, absolutely. And we'll see, I mean, the market, the market doesn't, someone will do it. So I want to get your thoughts on, let's go to the next level now. Okay, so web made things faster than the old analog world when digital, things are going to be faster with AI. Say the costs do come down. We live in a capitalist society. I'm sure going to happen with Amazon or with the competitors. Let's just say that it does go that next level. What happens next? Things are going to get faster. Information is going to be faster. The interface is conversational voice and maybe machine to machine. Everything's going on there. What's the future of our world? And I want to ask you this because you wrote a great essay on your newsletter on LinkedIn, the future of education that kind of teases out what the future might look like in a knowledge based world where as things get faster, antiquated and outdated becomes an issue. Even in state of the art systems, it's going to be a constant game of chasing that next real time value proposition. Absolutely. And so everybody's going to be rushing to adjust. I think all roads lead to what I call vertical AI, meaning I'm using the content and knowledge inside my organization and I'm making that available to more of my employees and more of my customers in a very efficient way, a way that I haven't been as efficient in the past. And so that's going to lead to what I call walled gardens of data and trying to protect the privacy of that data both from an IP standpoint and corporate strategy standpoint. This is going to start contrast to some of the big players like open AI, Google and Microsoft that really need access to a lot of content across the web. Enterprises aren't going to be so willing to share that and they're going to tap open to start opening up smaller models to unleash that knowledge inside their organization. So it's going to be a really wild next few years. It's going to be interesting. You mentioned walled gardens, my favorite word because if you go back 20 years, walled gardens were the antithesis of the open web. The web was about openness. And now you're starting to see walled gardens in the actual property of companies. Yet, the large language models are called proprietary. So instant flip script here. So what's your take on that? How do you rationalize that thing? And is that a good thing? And will this all interface with each other with data? So what's your angle on this? Well, on one hand it's a bad thing but on the other hand it's a very good thing for companies that want to compete on their knowledge and information. So if you want to compete on your knowledge and information, you got a lot of tools, lots of ways to do it. So I think it's going to create the needs to now selectively share that knowledge and information with partners and collaborators. So this machine to machine AI or collaborative AI as I call it is going to grow the ability to take your vertical AI and match it up with horizontal AI. We'll take chat GPT as an example, horizontal AI. You take chat GPT, you fine tune it or you use what we call prompt tuning to offset or augment it with your local content. Now you're starting to marry horizontal and vertical AI pieces. I think that's going to grow become, there might even be a brokering mechanism down the road. New information providers who can broker AI to AI to communications are going to be working. We're going to multi-agent AI systems. It's not going to be the prompt engineering of today. It's going to be the orchestration of many AI modules probably across multiple company boundaries. So great new opportunity for a whole lot of new firms still undetermined what that's going to look like. So yes. He's this out in your paper. I want to unpack that horizontal and vertical concept there because what you're saying is there's a horizontal layer called that domain stream. And then you get vertical in the domain expertise or metadata, if it's data, how do you explain this horizontal and vertical knowledge base architecture? Well, let's look at the combustion engine from the turn of the century hitting agriculture. That's horizontal general purpose technology that everybody uses. You really can't alter it. You can buy a different size combustion engine and you can do different things with it, but you can't alter it. Here, you can take the AI tools and turn it loose on your content and now it's altered to you. So that's a vertical sort of approach. If you're creating a new thing, essentially, that's part and parcel of your organization's information and knowledge. And so that's the horizontal and vertical. Horizontal would be like the combustion engine. Vertical would be the combustion engine completely redone just for you perfectly. That's awesome. I got to ask you, and again, back to your future of education piece, future of education horizontal and vertical AI and knowledge flow. I want to ask you something and I know you're in education, but I'll bring it up anyway because I threw a haymaker on one of our podcasts with Dave Vellante and I made a conjecture. And again, this is me on the podcast. So take it with a salt because we pontificate a lot. I said to Dave, I said, if you believe that the knowledge graph in the future is here, it's going to be the future, LMS is learning management systems of the old, very linear, very old school, slower, could be outdated. And I said the following statement. We are potentially on an educational crisis of the equivalent of the mortgage crisis where we're so overlevered with education investment, but yet the content is not adequate and yet could collapse. Now, again, education will collapse, but the thesis is if we can't get an architecture for content matching the needs of the student and you kind of tease, almost to say it's in the paper, but you're kind of teasing out that getting content at the right time is an imperative for educational institutions because there's a lot of money at risk and the capital markets right now from what I've been reporting is tight with education. There's a lot of need for potentially bailouts or leverage debts out there. So there's a lot of pressure in education. What's your reaction to that? Well, first of all, I'm going to put some things out here that are counter to that. The value of the education is tremendous in terms of lifelong earnings of the student who gets the degree. And that's now back to historic levels of highs from the 1920s. World War II witnessed a great leveling of access to education. We've reversed that over time. So whether you like education or not, the reality is you're going to get a million to $2 million of excess earnings with the college degree than you won't. So that's going to create tremendous demand no matter what happens with price points. The other piece of education, well, the learning management system can certainly be impacted. Long ago, it's already been altered a little bit. The learning management system is now a place to hook different modules that interoperate with each other. In fact, I'm on the board of One Ed Tech. That's what we specialize in, trying to get modules of interoperability embedded in an LMS. That's going to continue. I think we're going to see some really interesting stuff as in the books, the textbooks. Textbook manufacturers are going to make those generative AI things. And so they're going to be placed in many different places. You might see instructors adding to that in a certain way. I think we're in an evolution away from the industrial age technique of having the large lecture class to now having generative AI deliver more of that content and now have instructors work on motivation, connection, and meaning. The things that humans can do that generative AI cannot do. And so I'm hoping that that would be part of our future trajectory here in education. Great, great, and now thank you for sharing that great, great call out there. Let's go to that next level, since you brought that up. One of the topics around AI is, does it replace the humans? You kind of mentioned it earlier at the top. People aren't afraid of it. They shouldn't be. The motivation you mentioned about that, brings up this next role of what to do with the free time you have. If you're learning and or doing things with AI, that might've been mundane, toil, heavy, undifferentiated heavy lifting, as they say in the tech business. If AI can be an augmentation to the human, not just a chatbot, but like co-pilot-like augmentation, that implies value. So if that value is probably some of the boring toil or work that you can grind on with AI rather than the classroom, it frees up people's time. And so that's going to create some creativity and open opportunities. How do you see that happening or rolling out in terms of how do we deal with that? What fills the vacuum there? What fills that void? Well, for companies and educational everybody, starting at the top working down, you have to decide where to allocate resources from and to. So anytime you can allocate resources away from mundane, boring stuff and you can allocate it to more interesting, what I call value generating work, that can make a lot of people happy, including many of the staff. Now there's a certain collection of staff and you can perhaps count me as part of that, where we kind of like that boring mundane work and we don't like it to go away. But the creative destruction forces in the economy are going to weed that out quickly over time. Yes, there'll be some job displacement, but I think you're going to see much more of an augmentation effect, probably two thirds to three quarters augmentation effect rather than a replacement effect. To almost any job you can take a stick at. The message we give is, you're not going to be replaced by AI, but you may be replaced by somebody who can use AI. Being AI operator, it's interesting. It's the stakeholders remain the same. Yes, people still going to need to be educated. They're still going to need to find a job. They're going to still need mentoring and peer review or other services that you get to get out in the workforce. All those things are going to probably be augmented by AI. So that's kind of the key call out there. And that's going to be something that has to be a mind shift. So how do the institutions do that? Well, you know, learning is not a dry thing. It interacts with the human emotion and desire, the learner's desire. So institutions and educators really got to tap into the motivation. When you can double the motivation of the learner, you're going to quadruple their learning. The generative AI can help with a little bit of that motivation, but not all of it. And so institutions are going to have to now figure out how to enrich their pedagogy and their teaching so that they can use a generative AI, but then they can start to move some attention and effort to that deeply human element of learning. I want to get your thoughts quickly on the democratization angle and every of these big waves, this democratization access is one democratization. And talk about what needs to happen from a technology perspective, because this is not your yesterday's IT architecture. It's an AI system that's emerging. You're pushing buttons, you're turning knobs, you got personalization is a big part of AI. How do you see this being teed up or thought through from a holistic systems perspective? Well, the access component is kind of interesting because this is a class of technology that's kind of new for us really in the technology world that it helps the less skilled knowledge worker, more so than the highly skilled. It's really interesting. That's the man up and true for other, almost any other form of technology that we've had so far. The other side of that though is it's a ration to good right now is because of the pricing of the GPUs and chips that are out there and the services. So prices got to come down to get the democratization out there. For institutions, we got to make sure we make this free and affordable to institutions. I think any student who's got some money in their pocket are going to be buying and subscribing to AI services to help them in education. My thoughts go to the student who doesn't have the money in the pocket. I'm a former Pell Grant recipient and I was scraping by in my undergraduate curriculum. So I'm very sensitive to that and how to get that democratized. So I think institutions can certainly help a lot in that regard with their AI strategy and make sure they can address what I call the social mobility equity side of it. But as a larger public policy issue it's going to be a problem. I got to ask you about the motivation. I love how you brought that in. It keeps coming back where there's democratization from a motivation standpoint, getting that aspiration or a transition, an organizational. So as a leader in an organization I'm going to be thinking about, okay I want to make sure I don't foreclose the future but I don't want to overdrive the past with hype. I got to deliver the right sequence of operations here. How should I be thinking about if I'm an executive out there, how should I be thinking about how to get people enthusiastic and confident? So that motivations impact. How do you drive the motivation? Can you share your thoughts on that? Well, one thing is we've had more requests inside our organization for us to go out and explain this new technology and it's from a point of desire and good. Said no one ever about an enterprise system change like an ERP system or a manufacturing system change. Nobody calls me up and says tell me all the wonderful things about that. But this is the one where it is. So now that the first step is you got to tap into that motivation in the staff that's really interested and curious. You then have to allow the period of ideation to occur with your organization. Start to understand how this really works, where the benefits may be. Then you're gonna start to develop a whole bunch of use cases and then you start to prioritize use cases. And the use cases you want on board are the ones that are gonna give you the most confidence and feedback right away, easier to do, maybe smaller impact but shorter timeframe. And then start to build some successes and then start to work at bigger and bigger applications. I would not go for the most difficult value impactful thing right away. I would spend time to make sure you get your feet wet. Vince, I want to get your thoughts as we wind down here on a trend that we've been talking about a lot in theCUBE. And every cycle is always a fun thinking. Thinking, think like a designer. Iterate, data first, cloud first. There's a big focus right now we've been talking about this idea of systems thinking where there's consequences, not just isolating, building an app, thinking about the collateral damage of change and how to adapt. What's your view on systems thinking? Because you see a lot of people in this AI market, it's not just the young kids, it's the veterans who have systems architecture backgrounds. You're starting to see the guns come up as well. So you have kind of a diverse set of actors in this innovation curve. Talk about this idea of systems thinking. Do you see it out there? And if so, what mindset should people think about to be a systems thinking? Well, first off, I'm gonna give you my full feedback on what you said earlier. I compare the IT industry to the fashion industry. Every year a new fashion comes in and really goes, wow, the difference between us and the fashion industry, the fashion industry knows it's just fashion. In IT, we think it's real. However, I will say, yes, systems thinking or network thinking or complex systems thinking is absolutely in order. And first of all, generative AI is kind of that. It's a bit of a complex system in terms of how it works under the hood. And the probabilistic sort of framework you got to throw over your operations of AI now is very different. Like I said, this is not your father's data center anymore. This is a whole new thing. And so yes, you need broader holistic system thinking and complex system thinking and probabilistic system thinking skills. Business layer and the technical layer inside your organization. Vince, great to have you on. Final question, as you look at your journey, you've seen again, seen many ways of innovation and you've seen what's in front of us technically in the business, what do you hope to see happen? What's the path in your mind's eye that you hope to see a preferred future with generative AI? Well, I'm hoping it will certainly bring more knowledge to more people and make us all better, not just some of us. So the democratization aspects I'm very interested in. I think the disinformation and deep fake problem is gonna be hitting us. And so I'm a little worried about that as well. But in terms of the long-term future, certainly this is gonna help keep, stay too very competitive in terms of our business and industry and certainly in the global markets. The emerging markets that don't have an information economy they're gonna maybe miss this one out which is a bit of a concern. So the gap between the hasn't had nuts globally is probably gonna expand a bit. So like in all things like technology, there is always good but there's always some of the bad stuff too. You gotta let things rain. As Annie Grove would say, let chaos rain then rain in the chaos. You gotta keep in guard on it, keep on a leash so to speak. Vince, great to have you on and I appreciate your commentary and coming on and participating and contributing to our community on the SuperCloud 4th episode here about Gen of AI. Thank you. Thank you, John, this has been great. Awesome. Vince Callan, CIO of University of California in San Diego. He's on the front lines of education. He understands the system, mindset, the complex system and the opportunity that Gen AI brings. We have to be mindful of it, understand the value and implement it. It's an opportunity for everybody, lowers the barriers to entry in terms of value and knowledge and also levels everyone out. This is gonna be an opportunity we have to watch and we're gonna keep on watching it here in theCUBE. We'll be right back with more SuperCloud 4 after this short break.