 Okay, we're back here live in Las Vegas at Information on Demand, IBM's IOD, hash sign, IBM IOD. This is SiliconANGLE.com's exclusive coverage of information on demand here in Las Vegas. This is our flagship program, The Cube. We go out to the events to extract a signal from the noise. I'm John Furrier, the founder of SiliconANGLE.com and I'm joined by my co-host. I'm Dave Vellante of Wikibon.org. We're here with Steve Gold, who is the head of marketing for IBM Watson, room one. First of all, Steve, welcome to The Cube. Thank you, gentlemen. So, Dave, I think your mic might be out, but Steve, so thanks for coming on. Honestly, we're going to talk to you about Watson and big data, and especially everyone knows it's a big storyline you guys have had an amazing success with, with the program and the publicity's been phenomenal, but it really, it's not new to IBM, right? So, update the folks about the Watson situation, you know, how this is evolving. What's the current state of Watson? So, John, I think for the most part, people are very familiar with Watson and the Jeopardy situation and how it came to bear. And Jeopardy really represented a key inflection point. It brought together a core set of technological capabilities that had never really been demonstrated outside of the lab. Systems that were capable of both ingesting large amounts of data, big data, unstructured data. Systems that were capable of thinking and acting more like the human mind would than the binary logic of a chip. Systems that were actually capable of learning. So, Watson is a system that actually gets smarter with time and that opens up a whole new set of possibilities and capabilities. Jeopardy was clearly a great proof point of how that would work. But when we think about the possible ways in which Watson can be applied, any situation in which information is in abundance and accelerating, where, you know, time and accuracy to a response is critical. It tends to lend itself towards what many are calling a cognitive system, systems that are actually thinking. So, we began in late 2011 putting Watson to work in healthcare and we began with a number of specific use cases and I'm pleased to say we just literally went live from pilot to production with our first use case around the pre-authorization process. Many of us as patients are familiar. We go to the primary physician. They recommend perhaps an MRI or a CAT scan. That goes off to the plan provider for analysis and a response and approval. And today, that's a very manual process. Watson can be very objective. It can collect the longitudinal patient information. It can combine it with best practices and clinical guidelines and policies. And it can present to the clinician evidence-based insights as to whether or not that would in fact be the appropriate course of treatment. Let's take a step back. Let's define for the audience out there, what is Watson? And then let's go and look under the hood. So, what is Watson? For the one time, let's just get this definition on the record. So, Watson really represents a new class of industry-specific analytical solutions. So, if we break that down new class, it's a class of technologies based on this cognitive capabilities. It's industry-specific. If we think about what ultimately fuels these technologies, it's big data. So, is Watson a technology or a product? Well, I think Watson is an assemblage of technologies. It's not a single technology. And I think it manifests itself in the form of a product or a solution, if you will. And then the last thing is analytically based. So, what Watson does, it's a little bit different if you think about how to define that new class of industry-specific analytical solutions. What it does is it really brings together 41 subsystems of technology. And it allows an individual or a machine to input a question. And then it can literally, based on natural language, break down that question. Now, that's a pretty big feat in itself. Think about the human language. We say things like houses burn up as they burn down. Not very intuitive to a computer. Or why is it that feet smell but noses run? These are things that have historically created problems. And so, Watson, just in its natural language capability and navigating the complexities of human speech is pretty amazing. But the second thing it does, it literally can then break that down into all of its intended meaning. So it hypothesizes what is it that's really being asked and then it compares it to the possible responses. Let's talk about under the covers here because what you're really talking about is a reasoning capability. You're talking about a very complex system. And when we were at Intel's developer forum just this past couple of months ago, one of the things they talked about in the keynote was not embedded systems, intelligent systems. So obviously we're moving to a world where it's complex. So, obviously with big data, we've been covering startup scenes and kind of the big areas around obviously tsunami of data. Big, a lot of data coming into the world. Compute power is required to crunch even more of the data. So Moore's law is under under stress to do more. And then obviously it's impacting vertical markets. New applications are emerging in verticals like every vertical, oil and gas, financial, you name it, every vertical is impacted by big data. Okay, we got that. Okay, check. Check the boxes. This is an opportunity. So that's not simple, right? So what Watson's doing that's not in those three areas is it's using semantic processing, it's using natural language, machine learning. Correct. Explain to the folks that part of it. I mean, what's going on under the covers relative to, how does a computer reason? How does a computer figure this out? Is it Sol software? Is it hardware and software? All of the above? So it's a combination. So to your point, it's spot on on natural language. It's semantically and semantically breaks down an input. It analyzes that for the possible intentions or meanings. It does that through algorithms, so analytics. So think about a question being asked and it then looks through that vast repository of knowledge or information and it brings back a whole set of responses. And it then weights those responses with a degree of confidence. So think about this not as a deterministic outcome, but a probabilistic. Best example of that is if you ask a machine two plus two it responds with confidence today that it's four. And we'd all nod our heads as, of course, Watson would say, I'm confident four is an answer, maybe 90% confident, but four could also, or two plus two could also refer to a car configuration, two front seats, two back seats we're talking about automotive. It can refer to a family unit if we're talking about two parents and two children. We're in Vegas, two plus two is actually a poker strategy. So Watson is able to probabilistically bring back a set of responses and weight it. Very unique systems don't do that today. And then it's able to learn. So independent of the response that it generates and the confidence, the iterative nature of Watson says it gets smarter with time. You just took me to where I was going to go next, which is actually the software's hard to do. It's AI-like, it's very cool. The computational, the math involved is not trivial, requires a lot of computation, blades, et cetera. But then when you get to these vertical markets, they're different semantically. I mean, they have different lingo. Inglistics are different, how they talk about things. So you're talking about machines that are learning and evolving, adapting to the business. Take us through how IBM sees that, because that's a really important point because the diss on the computer industry is that you write some software and it's hard to port, get it re-write it. So having a learning machine is a really big deal. So talk about how that works and how that affects vertical markets in particular. Because they're different. You've got hedging, real-time financial trading, to oil and gas exploration. Absolutely, and it's the industry that brings, obviously, the degree of confidence to the application. And so if you look at the work we're doing in healthcare, we've partnered, for example, Memorial Sloan Kettering, probably the preeminent institution for cancer. And we're really drawing upon their expertise. And so what IBM is really bringing forward isn't just Watson Technologies, it's the combination of the big data capabilities we have and the Watson Technologies, combined with the infrastructure that it's running on. All of this has to come together with the expertise of the industry. And it's really a collaborative effort that's allowing us to bring these solutions forward. So we're here with Steve Gold, who's the head of Worldwide Marketing for IBM Watson. Steve, when IBM Deep Blue beat Gary Kasparov, that was sort of binary, and it was kind of, after that, it was a little bit anticlimactic for the outside observer. I remember hearing you talk about the progression that Watson made in terms of getting to the level of a Jeopardy! Grandmaster. And it took some time, it took some iteration. It did. But it was actually, that ascent was relatively fast in the grand scheme of things. I know there's a lot of hard work behind it. You've achieved that. Is the progression that Watson's making now is ascendancy continuing at that trajectory? How do you measure that? What's it look like today? Yeah, great question. So I think if you look at, it took research, really four and a half years to incubate what is an amazing technological outcome. And from that, we've gone from that incubation through Jeopardy! to today, in a production instantiation of a Watson solution in a year. So we've gone from four and a half years in research to the first production instance in a year. I fully believe it will continue to accelerate, but we're still in its infancy, right? It's still young, we're still learning. I think as John points out, there's still a lot of work to do with natural language and ontologies and building up taxonomies and annotation that has to occur. So there's still a fair amount of investment, if you will, that's required. To continue to advance this from a commercialized point of view. And that will continue to go, clearly just accessing information and making sense of that information itself is daunting, let alone advancing the use of its application and the algorithms that have to support the particular application that we're building. So I think you'll see a steady cadence of both applications coming forward, as well as technological advancements. I mean, we continue to work very closely with research, right? So that work continues on and I think it'll manifest itself in the form of new capabilities and new ways to deliver and new ways to interact with a Watson. So the Watson of commercialization, it's not like you're going to drop in a block of Watson to a use case necessarily. Perhaps there is one like that, but so pieces of Watson will get commercialized. Is that the right way to think of it? Well, I mean, Watson is a single piece, if you will. It certainly represents, if you want to draw the analogy to a car, I could dismantle a car. Each of the pieces that I would dismantle doesn't make a car. When I put them together in a certain prescriptive way, I then am able to achieve something that is a desired effect. And that's exactly what Watson is. It's the end result of bringing together disparate technologies in a very formulaic way to accomplish a very purpose-built outcome. In this case, if you describe it, it's the ingestion of big data, disparate data, unstructured data, it's understanding and reasoning, it's presenting back evidence-based outcomes as weighted probabilistic-based outcome, and then it's learning. And that's really the end effect of what Watson could do. Clearly applicable to healthcare, we're moving into financial services. We announced last March with our partnership with Citi. It's muted on that. And then we'll continue to advance a steady cadence, both by industry and use case. So the model actually will be to buy a Watson solution and apply it and mold it to a particular industry. Absolutely, I think today that's the case, because again, we have to ingest relevant data, and that will be by industry and use case. As we grow the corpora of information that we have available, you could start to think of more broad horizontal type applications. We're often asked the question, why can't Watson, do you speak, queried about general medicine? And it's because today, Watson has a very prescriptive set of insights into medical literature and clinical studies and patient data that's aligned to particular diseases. But as you accumulate that information, the ability to query Watson for a broader set of interests is certainly relevant. Now, Steve, you're a startup guy. You've done a number of startups, you've sold some companies, you've run a publicly traded company, I believe. How is what you're doing now inside of IBM similar and how is it different? Yeah, so it's really interesting. IBM obviously has some tremendous minds. In fact, I would profess it's probably the only organization that could have brought together the technologies and the people to bring a Watson to market. So the depth of resources unparalleled, the resource relative to advancing and commercializing this is a formidable undertaking. And in fact, many ways, Watson looks exactly like a startup in terms of how it's coming to market, of how it's being incubated and advanced with the benefit of having obviously the full support of the IBM company. Obviously, it was a little bit different in that context, is this is a little bit of green field, right? IBM has been a bellwether in terms of the technologies that it's brought to market. And so we're pioneering really a whole new era of computing, very exciting era of computing that in totality, I don't think an individual startup would be able to do. Yeah, we think, we think, we've been talking, I'll show you what you're all, we're on the big data religion. We think that the impact is so transformative, we think it's like the PC revolution and client server combined and happening at a much accelerated rate with big data. And it's exciting. So early, it's hard to do. But my question is using big data. So you're taking Watson from a marketing stunt with the jeopardy thing that was cool. To me, that's like the tech version of the space jump, right? It's like, you know, it's like big data in action on jeopardy, it's fantastic. But now you're putting it into practice. So the question is, how do people use Watson today? How can someone use Watson, both a large enterprise and say a small, medium-sized enterprise? Because that's the big question we're getting from folks is I want big data, especially small, medium-sized enterprises. How can I start using this stuff? So I think there's two questions. You know, one is can organizations start to put big data to work, right? And then what role does Watson play as it relates to big data? And I think the first question, I mean, I think organizations are just now awakening to the opportunities to take in mine information for better outcomes, for better action. Some of that is based on, you know, classical analytical technologies from business intelligence to predictive, right? So you don't necessarily need a Watson to solve all the problems. And that's an important differentiator is there's a lot of approaches to using big data. And I think about big data as kind of the fuel in the tank, right? So if we go backwards, big data is oil in the ground by itself, it's not all that interesting. When we start to refine it, we start to get, you know, more interesting outcomes and applications for how to use big data. Watson represents, I think, a unique application of use of big data in that it's iterative, it's stateful, it's a more natural way to interact with big data. So I can report against big data, get dashboards and scorecards, very useful about what's happened and why. I can predict big data about where we're going and look at those outcomes. But Watson is the here and now, right? Knowing what I know, what decision would I make that would be in the best interest of my client, my partner, my stakeholders. And so Watson's very complimentary to, I think, many of the initiatives organizations already have underway with big data. I think for the most part, we all understand big data is transformative, you know, but it's transformative in the context of how it's being applied, not in independent of itself. It's not just that I got big data and now I make better decisions, it's big data plus analytics, plus one of the things that I think we often gloss over is the need, right? If speed and accuracy aren't critical and I have plenty of bandwidth to sift through the information, then maybe the application of analytics and the need isn't there. That's not the case. Organizations are pressed to make timely decisions, that the nature of those decisions has amplified in criticality, so there's more at stake. They're pushing more out the line of business. And I think all of this is lending itself to that perfect storm of big data, plus analytics, plus a compelling need in the market, makes for some interesting times and opportunities for organizations. Steve, you used the fuel and the tank analogy. How is the fuel and the big data tank changing? And how is Watson responding to the different data sources? Can you talk about that and how it approaches that variety, if you will, of data sources? They have 90% of the world's data, and we've heard this has been created in the last two years. 2.5 quintillion bytes of information is created every single day. When you look at that, and you look at what that means, it says volume is almost... It's up to the right. Uncomprehensible, right? So you have this big volume of data. 80% of that data is unstructured. We've never really dealt with unstructured information. Traditional programmatic systems like rows and columns, right? So now I have all of this unstructured data and it's all disparate formats, right? So it's not like it's just sitting in one repository that I can clean up and utilize. I now need to pull from disparate sources. You know, I need to make sense of that information. And so that fuel and the tank has taken on a very new form that most organizations are ill-equipped to really maximize and put to work. Think about the proliferation of social data, right? There's over 250 billion emails sent every day. In those emails, probably locked away nuggets of insights about clients or partners or opportunities. 17 terabytes of tweets created every single day, right? What can we glean from those tweets that may be of interest to the behavior of our audience or our market? So I think the fuel, if you will, is very different today. That's a passionate subject to us, by the way, the whole tweet thing, because it's real-time people talking and like you can monitor it. And you can linguistically go at it and figure out things. Well, in fact, everybody says, well, I'm not really sure how a tweet would relate to, for example, healthcare. We said, well, what if there was an outbreak of ragweed? It was early pollen season. And I go to my doctor and the initial reaction is, you probably have the cold or the flu, but if there was advanced insight into the fact that people were tweeting about, hey, it's an early allergy season. And ragweed seems to be terrible this year. Signal. That's exactly right. It's a signal and that information can be mined, can be used, can be applied, and now that clinician has one more point of evidence that they can point to and make a decision. That's a great example. So we talked on theCUBE before about some big data applications or benefits, and one is, obviously, answering older questions faster, which, you know, with compute and more data, yeah, you go faster. So, you know, five minutes, five seconds, five days, five weeks, whatever, five hours. But it's the ability to answer new questions. So I want to get your perspective on that, because that really is a mind-blowing concept that questions that could never be asked before by businesses and individuals are now possible. They are, and actually, to put a twist to that, which, because that in and of itself is fascinating, but imagine now a system that's able to prompt back and ask questions, right? So one of the byproducts of Watson is this whole notion that in theory, right, it's not just the input in, but it's the recognition that if you would provide me with some additional insight, I could actually refine, right, my response with a higher degree of confidence, right? So knowing that you have, you know, the sniffles and the sore throat is important. It'd be great to know if you have a fever, right? So Watson's able not to think on its own, it's not conscious, there's no self-awareness, but what it's doing, it's using the evidence from the big data to say, hey, I know I have information out here that could be applied if I could fill in some of the missing voids. My final question for you, we're here with Steve Gold, talking about great stuff about Watson, but my final question's a little bit different, it's about people. And I really want to ask more about internal to IBM. IBM, as we've been covering you guys, has a lot of great talent. PhDs, the labs team, and R&D's phenomenal. There's a lot of stuff going on at IBM that people don't usually know about, so I want to ask you this. Big data is like a geek revolution, right? I mean, we're kind of geeking out here, just thinking about the provocative possibilities of big data on a business and user basis. What's it like at IBM right now for all the alpha geeks inside the company? Just share a perspective about what's the vibe like? I mean, they're doing handstands, it's like big data, like getting everyone all energized. What's some of the, share some anecdotal comments and observations? Yeah, I think, I would define it as re-energized. Right, I think what we have seen, IBM has been marching on a 2015 roadmap, and that roadmap has really been underpinned by a couple of key concepts, big data being won, analytics being won, and as the play has evolved, I think people have just been in awe and of the fascination that says, wow. You know, we worked for a company that was well ahead of the curve, right? They saw this coming, they put it to work, and I'm part of that revolution, I'm part of making a difference, I'm part of advancing technology in a way that truly can impact society. I mean, you know, if you think about smarter planets, smarter planets, not about technology, right? It's not about a solution, it's not about an application. Smarter planet is altruistic, it's about how do we improve the quality of our lives. Big data is a big part of that, right? The best technology should be invisible. It should, it's completely transparent, right? And in fact, you know, one of the things that we just talked about the other day is the whole user experience around Watson, right? How does it alter that the keyboard, right, monitor in a type experience that we've grown up with? If I can interact with a system that's intelligent, that's learning, that's able to be stateful and understand where I'm going and what I'm asking, does that redefine, you know, in totality, how I work with systems, and I think it'd be fascinating. I think, you know, I'm excited about what you're doing. Steve Gold on theCUBE, we have to break and come out next segment. I just want to say I'm really excited about what you guys are doing with Watson. You know, I clicked for me in this interview that the Watson-Jeopardy moment was like the space jump in terms of like the big data, you know, mind-blowing kind of capabilities. So congratulations, really big effort and you're bringing it into reality. Congratulations, IBM's got some exciting stuff and all kinds of great tech with Watson and beyond. This is theCUBE, it's looking at angles flagship program and got the events to extract the signal from the noise. We'll be right back with our next guest after this short break.