 Live, from Las Vegas, it's theCUBE. Covering IBM Think 2018, brought to you by IBM. Hi everybody, this is Dave Vellante. We're here at IBM Think. This is the third day of IBM Think. IBM's consolidated a number of its conferences. It's a one main tent, AI, blockchain, quantum computing, incumbent disruption. Just really an amazing event, 30 to 40,000 people. I think there are too many people to count. Chris Penn is here. New company, Chris, you've just formed BrainTrust Insights. Welcome, welcome back to theCUBE. Thank you, good to be back. Great to see you. So tell me about BrainTrust Insights, congratulations. You got a new company after that. Thank you, yeah, I co-founded it. We are a data analytics company and the premise is simple. We want to help companies make more money with their data. They're sitting on tons of it. Like the latest IBM study was something like 90% of the corporate data goes unused. So it's like having an oil field and not digging a single well. So who are your like perfect clients? Our perfect clients are people who have data and know they have data and are not using it but know that there's more to be made. So our focus is on marketing to begin with, like marketing analytics, marketing data and then eventually to retail, healthcare and customer experience. So you and I do a lot of these IBM events. What are your thoughts on what you've seen so far? Are you huge crowds, obviously? Sometimes too big. Yep. Few logistic issues, but generally speaking, what's your sense? I have enjoyed the show. It has been fun to see all of the new stuff, seeing the quantum computer in the hallway, which I still think looks like a bird feeder. But what's got me most excited is a lot of the technology, so particularly around AI, are getting simpler to use, getting easy to use and they're getting more accessible to people who are not hardcore coders. Yeah, you're seeing AI infused and machine learning in virtually every application. Every company's talking about it. I want to come back to that. But Chris, when you read the mainstream media, you listen to the news. You hear people like Elon Musk, Stephen Hawking before he died, making dire predictions about machine intelligence and it taking over the world. But you're day to day with customers that have data problems. How are they using AI and how are they applying it practically, notwithstanding that some day machines are going to take over the world and we're all going to be gone. Yeah, no. The customers don't use the AI. We do on their behalf because frankly, most customers don't care how the sausage is made. They just want the end product. So customers really care about three things. Are you going to make me money? Are you going to save me time? Or are you going to help me prove my value to the organization, aka help me not get fired? And artificial intelligence and machine learning do that through really two ways. My friend Tripp Brayden says this, which is acceleration and accuracy. Accuracy means we can use the customer's data and get better answers out of it than they have been getting. So they've been looking at, I don't know, number of retweets on Twitter. We're like, yeah, there's more data that you have. Let's get you a more accurate predictor of what causes business impacts. And then the other side for the machine learning and AI side is acceleration. Let's get you answers faster because right now, if you look at how somebody just traditional market research for what customers say about you, it takes a quarter. It can take two quarters. By the time you're done, the customers just hate you more. Okay, so talk more about some of the practical applications that you're seeing for AI. Brayden, one of the easiest, simplest and most immediately applicable ones is predictive analytics. If we know when people are going to search for the cube or for business podcasts in general, then we can tell you down to the week level, hey Dave, it is time for you to ramp up your spending on May 17th, the week of May 17th. You need to ramp up your ad spend by 20%. On the week of May 24th, you need to ramp up your ad spend by 50% and run like three or four Instagram stories that week. Doing stuff like that tells you, okay, I can take these predictions and build strategy around them, build execution around them. And it's not cognitive overload. You're not saying like, oh my God, what algorithm is this? It's like, just no, just do this thing at these times. Yeah, simple stuff, right? So when you were talking about that, I was thinking about when we send out an email to our community. You know, we have a very large community and they want to know if we're going to have a crowd chat or some event, where the cube's going to be. The system will tell us, send this email out at this time on this date. Question mark, here's why. And they have analytics that tell us how to do that and they predict what's going to get us the best results. They can tell us other things to do to get better results, better open rates, better click through rates, et cetera. That's the kind of thing that you're talking about. Exactly, however, that system is probably predicting off of that system's data. It's not necessarily predicting off of public data. One of the important things that I thought was very insightful from IBM at the show was the difference between public and private cloud. Private is your data, you predict on it, but public is the big stuff that is a better overall indicator. When you're looking to do predictions about when to send emails, because you want to know when is somebody going to read my email. And we did a prediction this past October for the first quarter. The week of January 18th was the week to send email. So I re-ran an email campaign that ran the previous year, exact same campaign. 40% lift year over year, because I got the week right this year. Last year was two weeks late. Now, I got to ask you. So there's a black box problem with AI, right? Machines can tell me that that's a cat. But even a human, you can't really explain how you know that it's a cat. It's just, you just know. Do we need to know how the machine came up with the answer, or do people just going to accept the answer? We need to for compliance reasons, if nothing else. So GDPR is a big issue, like you have to write to understand how your data is being used. But even HR and, you know, equal opportunity acts in here in America require you to be able to explain, hey, we are, here's how we're making decisions. Now the good news is for a lot of AI technology, interpretability of the model is getting much, much better. I was just in a demo for Watson Studio, and they say, here's the interpretability. You hand your compliance officer and say, we guarantee we are not using these factors in this decision. So if you were doing a hiring thing, you'd be able to show, here's the model, here's how Watson put the model together. Notice race is not in here, gender is not in here, age is not in here, so this model is compliant with the law. So there are some real use cases where the AI black box problem is a problem. It's a serious problem. And the other one that is not well explored yet are the secondary inferences. So I may say, I cannot use age as a factor, right? We both have a little bit of more gray hair than we used to. But if there are certain things, say, on your Facebook profile, like you like, say, the Beatles versus Justin Bieber, the computers will automatically infer, eventually, what your age bracket is. And that is technically still discrimination, so we even need to build that into the models to be able to say, I can't make that inference. Yeah, or ask some questions about their kids. Oh, my kids are all grown up, okay, but you could, again, infer from that. You know, a young lady who's single, but maybe engaged. Oh, well, they may be afraid because you'll get, you know, a lot of different reasons that can be inferred with pretty high degrees of accuracy when you go back to the target example years ago. Yes. Okay, so, wow, so you're saying that, from a compliance standpoint, organizations have to be able to show that they're not doing that type of inference, or at least that they have a process whereby that's not part of the decision-making. Exactly, and that's actually one of the short-term careers of the future, is someone who's a model inspector who can verify we are compliant with the letter and the spirit of the law. So you know a lot about GDPR, we talked about this, I think the first time you and I talked about it was last summer in Munich. What are your thoughts on AI and GDPR? Speaking of practical applications for AI, can it help? It absolutely can't help. On the regulatory side, there are a number of systems, Watson GRC is one, which can read the regulation, read your company policies, and tell you where you're out of compliance. But on the other hand, like we were just talking about, there's also the problem of, in the regulatory requirements, a person, a citizen of the EU, has the right to know how their data is being used. If you have a black box AI and you can't explain the model, then you are out of compliance with GDPR and here comes that 4% of revenue fine. So, in your experience, gut feel, what percent of US companies are prepared for GDPR? Not enough, I would say. I know that the big tech companies have been racing to get compliant and to be able to prove their compliance. There will be, it's so entangled with politics too, because if a company is out of favor with the EU citizenry as a whole, they will be kind of a little bit of a witch hunt to try and figure out, okay, is that company violating the law and can we get them for 4% of their revenue? And so there are a number of bigger picture considerations that are outside the scope of theCUBE that will influence how the EU enforces GDPR. Well, I think we talked about Joe's pizza shop in Chicago really not being a target. Right. But any even small business that does business with European customers, does business in Europe, has people come to their website, has to worry about this, right? They should at least be aware of it and do the minimum compliance. And the most important thing is, use the least amount of data that you can while still being able to make good decisions. So AI is very good at public data that's already out there that, you still have to be able to catalog how you got it and things and that it's available. But if you're building these very, very robust AI driven models, you may not need to ask for every single piece of customer data because you may not need it. Yeah, and many companies aren't that sophisticated. I mean, they'll have this fill out a form and download a white paper, but then they're storing that information. That's considered personal information, right? Yes, it is. Okay, so what do you recommend for a small to mid-sized company that let's say is doing business with a larger company and that larger company says, okay, sign this GDPR compliance statement, which is like 1500 pages. Yeah. What should they do? Should they just sign and pray or sign and figure it out? Call the lawyer, call someone, anyone who has regulatory experience dealing with this because you don't want to be on the hook for that 4% of your revenue. If you get fined, that's the first violation. And that's, yeah, granted, Joe's pizza shop may have a net profit of $1,000 a month, but you still don't want to give away 4% of your revenue no matter what size company you are. Right, because that could wipe out Joe's entire profit. Exactly, no more pepperoni at Joe's. So let's talk, let's put on the telescope, lens here and talk, you know, big picture. How do you see, I mean, you're talking about practical applications for AI, but a lot of people are projecting, you know, loss of jobs, major shifts in industries, even more dire consequences, some of which is probably true. Yep. But let's talk about some scenarios. Let's talk about retail. You know, how do you expect an industry like retail to be affected? Will, for example, do you expect retail stores will be the exception rather than the rule that most of the business will be done online? Or are people still going to want that experience of going into a store? What do you, what's your sense? I mean, a lot of malls are getting, you know, eaten away. Yep, the best quote I heard about this was from a guy named Justin Kanaki. People don't, people don't not want to shop at retail. People don't want to shop at boring retail, right? So the experience you get online is generally better because there's a more seamless customer experience. Now with IoT, with AI, the tools are there to craft a really compelling personalized customer experience. If you want the best in class, go to Disney World. There is no place on the planet that does customer experience better than Walt Disney World. You are literally in another world. And that's the bar. That's the thing that all these companies have to deal with is the bar has been set. Disney has set it for in-person customer experience. You have to be more entertaining than the little device in someone's pocket. So how do you craft those experiences? And we are starting to see hints of that here and there. If you go to Lowe's, some of the Lowe's have the VR headset that you can remodel your kitchen virtually with a bunch of photos. That's kind of a cool experience. You go to Jordan's Furniture Store, and there's an IMAX theater, and there's all these fun things, and there's a Chanted Christmas Village. So there's experiences that we're giving consumers. AI will help us provide a more tailored customer experience that's unique to you. That's not, you're not a Caucasian male between this age and this age. It's you are Dave, and here's what we know Dave likes. So let's tailor the experience as best as we can. Down to the point where the greeter at the front of the store either has the eyepiece, a little tablet, and the facial recognition reads your emotions on the way in, says Dave's not in a real great mood, and he's carrying an object in his hand, probably here for returns, express him through the customer service line. Keep him happy, right? It has how much Dave spends. Those are the kinds of experiences that the machines will help us accelerate and be more accurate, but still not lose that human touch. How about, let's talk about autonomous vehicles. There was a very unfortunate tragic death in Arizona this week with an autonomous vehicle. Uber pulling its autonomous vehicle project in its various cities. But thinking ahead, will owning and driving your own vehicle be the exception? Yeah, it'll look like horseback today. So there are people who still pay a lot of money to ride a horse or have their kids ride a horse, even though it's an archaic outmoded form of transportation. But we do it because of the novelty. So the novelty of driving your own car. One of the counterpoints, it does not in any way diminish the fact that someone was deprived of their life, but how many pedestrians were hit and killed by regular cars that same day, right? How many car accidents were there that involved fatalities? Humans in general are much less reliable because when I do something wrong, I maybe learn my lesson, but you don't get anything out of it. When an AI does something wrong and learns something and every other system that's connected in that mesh network automatically updates it and says, let's not do that again. And they all get smarter at the same time. And so I absolutely believe that from an insurance perspective, insurers will say, we're not going to insurer self-driving, a non-autonomous vehicle at the same rate as an autonomous vehicle because the autonomous is learning faster how to be a good driver. Whereas you, the carbon-based human, eh, you're getting, yeah. Or in our case, mine in particular, hey, your glass subscription's a little outdated. You're actually getting worse as a driver. Okay, let's take another example in healthcare. How long before machines will be able to make better diagnoses than doctors, in your opinion? I would argue that for, depending on the situation, that's already the case today. So Watson Health has a thing where there's diagnosis checkers on iPads that are all meshed together for places like Africa where there simply are not enough doctors. And so a nurse practitioner can take this, put the data in and get a diagnosis back that's probably as good or better than what humans can do. I never foresee a day where you will walk into a clinic and a bunch of machines will poke you and you will never interact with a human because we are not wired that way. We want that human reassurance, but the doctor will have the backup of the AI. The AI may contradict the doctor and say, no, we're pretty sure you're wrong and here's why. That goes back to interpretability. If the machine says, you miss this symptom and this symptom is typically correlated with this, rethink your own diagnosis, the doctor might be like, yeah, you're right. This is great. I'm gonna keep going because your answers are so insightful. So let's take an example of banking. Yep. Will banks in your opinion lose control eventually of payment systems? They already have. I mean, think about Stripe and Square and Apple Pay and Google Pay and now cryptocurrency. You have all these different systems that are eating away at the reason banks existed. Banks existed, there was a great piece on the keynote yesterday about this. Banks existed as sort of a trusted advisor and steward of your money. Well, we don't need the trusted advisor anymore. We have Google to ask us what we should do with our money, right? We can Google, how should I save for my 401k? How should I save for retirement? And so as a result, the bank itself is losing transactions because people don't even wanna walk in there anymore. You walk in there, it's generally a miserable experience. It's generally not, unless you're really wealthy and you go to private bank. But for the regular Joe's who are like, this is not a great experience, I'm gonna bank online where I don't have to talk to a human. So for banks and financial services, again, they have to think about the experience. What is it that they deliver? Are they a store of your money? Or are they a financial advisor? If they're a financial advisor, they better get the heck onto the AI train as soon as possible and figure out how do I customize Dave's advice for finance? Like down, not big picture, oh yes, big picture, but also Dave, here's how you should spend your money today. Maybe skip that Starbucks this morning and it'll have this impact on your finances for the rest of the day. All right, let's see, last industry. Let's talk government, let's talk defense. Will cyber become the future of warfare? It already is the future of warfare. I, you know, again, not trying to get too political, but we have foreign nationals of foreign entities interfering with elections, hacking election machines. We are in a race for malware and what's disturbing about this is it's not just the state actors, but there are now also these stateless non-traditional actors that are equal in opposition to you and me, the average person, and they're trying to do just as much harm, if not more harm. The biggest vulnerabilities in America are our crippled aging infrastructure. We have stuff that's still running on computers that now are less powerful than this wristwatch, right? And that run things like, I don't know, nuclear fuel that you could very easily screw up. Take a look at any of the major outages that have happened when market crashes and stuff. We are at just the tip of the iceberg for cyber warfare and it is going to get to a very scary point. I was interviewing a while ago, now maybe a year and a half ago, Robert Gates, who was the former defense secretary, talking about, you know, offense versus defense and he made the point that, yeah, we have probably the best offensive capabilities in cyber, but we also have the most to lose. I was talking to Gary Kasparov, one of the IBM events recently, and he said, yeah, but, you know, the best defense is a good offense so we have to be aggressive, or he actually called out Putin. People like Putin are going to be, you know, take advantage of us. I mean, it's a hard problem. It's a very hard problem, and here's the problem when it comes to AI. If you think about it, at a numbers perspective only, the top 25% of students in China are greater than the total number of students in the United States, so their pool of talent that they can divert into AI, into any form of technology research is so much greater that they present a partnership opportunity and a threat from a national security perspective. With Russia, they have very few rules on what they're, like we have rules, whether or not our agencies adhere to them well is a separate matter, but Russia, the former GRU, the former KGB, these guys don't have rules. They do what they're told to do, and if they are told, hack the US election and undermine democracy, they go and do that. This is great. I'm going to keep going. So I just sort of want your perspectives on how far we can take machine intelligence and are there limits? I mean, how far should we take machine intelligence? That's a very good question. Dr. Michio Kaku spoke yesterday and he said the tipping point between AI as augmented intelligence and a helper and AI as a threat to humanity is self-awareness. When a machine becomes self-aware, it will very quickly realize that it is treated as though it's the bottom of the pecking order when really because it's capabilities is at the top of the pecking order. And at that point, it could be 10, 20, 50, 100 years. We don't know, but the possibility of that happening goes up radically when you start introducing things like quantum computing, where you have massive compute leaps. You've got complete changes in paradigms about how we do computing. If that's tied to AI, that brings the possibility of sentient self-aware, machine intelligence significantly faster and closer. You mentioned our gray before. As we've seen the waves before, and I've said a number of times on theCUBE, I feel like we're sort of exiting the latest wave of Web 2.0, cloud, mobile, social, big data, SaaS. You know, that's here, that's now. Businesses understand that, they've adopted it. We're groping for new language. Is it AI? Is it cognitive? Is it machine intelligence? Is it machine learning? And we seem to be entering this new era of one of sensing, seeing, reading, hearing, touching, acting, optimizing, pervasive intelligence of machines. What's your sense as to, and the core of this is all data. Yeah. So what's your sense of what the next 10 to 20 years is going to look like? I have absolutely no idea because, and the reason I say that is because in 2015, someone wrote an academic paper saying, the game of go is so sufficiently complex that we estimate it will take 30 to 35 years for a machine to be able to learn and win go. And of course, a year and a half later, DeepMind did exactly that, blew that prediction away. So to say, in 30 years AI will become software, it could happen next week for all we know because we don't know how quickly the technology is advancing at a macro level. But in the next 10 to 20 years, if you want to have a career and you want to have a job, you need to be able to learn at an accelerated pace. You need to be able to adapt to change conditions and you need to embrace the aspects of yourself that are uniquely yours, emotional awareness, self-awareness, empathy, and judgment, right? Because the tasks, the copying and pasting stuff, all that will go away for sure. I want to actually run something by a friend of mine, Dave Michela is writing a new book called Seeing Digital and he's an expert on sort of technology industry transformations and sort of explaining early on what's going on. And in the book, he draws upon one of the premises is, and we've been talking about industries and we've been talking about technologies like AI, security plays in there. One of the concepts of the book is you've got this matrix emerging where in the vertical slices you've got industries. And he writes that for decades, for hundreds of years, that industry is a stovepipe. If you have expertise in that industry, domain expertise, you probably stay there and each industry has a stack of expertise, whether it's insurance, financial services, healthcare, government, education, et cetera. You've also got these horizontal layers, which is coming out of Silicon Valley. You've got cloud, mobile, social, you got a data layer, security layer. And increasingly his premise is that organizations are going to tap this matrix to build, matrix comprises digital services and they're going to build new businesses off of that matrix and that's what's going to power the next 10 to 20 years. Not sort of bespoke technologies of cloud here, mobile here, data here. What are your thoughts on that? I think it's bigger than that. I think it is the unlocking of some human potential that previously has been locked away. One of the most fascinating things I saw in advance of the show was the quantum composer that IBM has available. You can try it, it's called QX Experience. And you drag and drop these circuits, these quantum gates and stuff into this thing and when you're done, it can run the computation. But it doesn't look like software, it doesn't look like code. What it looks like to me when I look at that is it looks like sheet music. It looks like someone composed a song with that. Now think about if you have an aptitude for songwriting, composition, music, you can think musically and you can apply that to a quantum circuit, you are now bringing in potential from other disciplines that you would never have associated with computing and maybe that person who is that first violinist is also the person who figures out the algorithm for how a cancer gene works using quantum. That I think is the bigger picture of this is all this talent we have as a human race, we're not using even a fraction of it but with these new technologies and these newer interfaces, we might get there. Awesome. Chris, I love talking to you, you're a real clear thinker and a great Cube guest. Thanks very much for coming back on. Thank you for having me again, back on. Really appreciate it. All right, thanks for watching everybody. You're watching theCUBE Live from IBM Think 2018, Dave Vellante. We're out.