 Hi, I'm John Furrier with SiliconANGLE theCUBE. We're here live in San Francisco covering the GE data forecast event. I'm with Michael Jordan from Berkeley. You see Berkeley on the panel with GE. Welcome to the interview. Thank you. So we are talking about data. GE is obviously progressive. Big data is a big part of their business machines. It's part of industrial internet. You know, business value is their shtick, which is great. You know, business value is great. But at the end of the day, data is a big part of today's culture. And I want you to comment from a tech perspective, some of the real changes around data. And not from a business value standpoint, that's a different conversation, but like true transformation in society and in computer science, because it really is a perfect storm between computer scientists going to this next level. What's your take on that? Because there's a lot of social change going on right now. What's your take on that? Great question. I think a lot of the work in computer science has been via database people who've had a very different perspective on inferential questions. So, you know, a database person, if you take a database class, you learn about storing people's home address, their age, their height, their income. And that data is the real stuff that you want to retain, you want to transform, and you want a query. Statisticians think differently about it. They want to make a predictive model based on the data. So, if someone comes in and has a certain profile, they want to predict whether they'll click on an ad or not, or predict whether they'll have a disease or not. And the people who form the original data may be dead and gone. So, you know, trying to understand the uncertainty that propagates by collecting data in a certain way, trying to understand the sampling pattern collecting a certain way is a very non-database kind of perspective, and it's what's needed to sort of go from where computer science has been to where it needs to go. Data is a double-edged sword. You know, in my career, in every big company I work for, data is like justifies decisions, and you can make data do anything. So, the question I have for you is, what is really the true value of the data from a database perspective? Do you think about the database first or the data? And the question we were talking about last week in Boston for an event we were at was, usually in the old days was, pick a database and then everything else gets done. Now it's not so much so. Now it's the data, and the database kind of falls into the background. What's your take on this? So, it's not the data or it's the population from which the data were sampled, and you're trying to ask questions about the population, because your new customers will come from the population, they won't come from the customers you've already seen. And so, in many situations, you might have a huge amount of data about Michael Jackson, Michael Jordan, or whoever, and you don't have very much data about so-and-so. So, the really interesting questions if you want to serve everybody is how do you take the strength and the data about so few individuals and percolate it to make strong predictions for other people? And you know, the principles on how to do that are not clear yet. Just a social, if you're my friend, you're like me, well that doesn't tend to work out very well. What other ways of connecting people and allowing predictions to flow in social graphs are really going to work, that's still pretty much open. But these are new data flows, though. This is a new kind of data. Is there any inhibitors in preventing the accelerated development of this? Well, false positives of all kinds are an inhibitor. So, in some areas, it doesn't give, you know, say Amazon, you're making recommendations. If you give me false positive, I'm just a little annoyed that you could recommend and gone with the wind to me, or if I bought a refrigerator, once you keep recommending refrigerators, that's a little annoying, but it's no big deal. But, and- It's annoying, like re-targeting's a pain in the ass. It's a pain in the ass. I mean, everyone likes re-targeting. I don't want to see the ad anymore. It's going to go away eventually, they're going to get smart about that. But still, there are other domains where, like medical decisions, you make a false positive. You tell me I got into some disease, which I really don't. So now I have to do an expensive test or a dangerous test to go forward, or I might, you know, make a bad decision. Well, I'm excited to talk to you. One of the things you mentioned on the panel here in San Francisco was, make a couple of comments. One was arithmetic is a gone thing, one plus one, it's not about one plus one anymore. It's more about Bayesian theory or other things. And that strikes a chord, because as a father and as young kids going into the college, for instance, a new generation is coming into the workforce and certainly as a parent. You know, I'm trying to figure out how to mentor the young generation. So I'd like to get your perspective on that. As kids grow up, whether they're in middle school, high school or going into college, what's your advice to them to look at the tooling and their mindset of this new world of, I'm going to say data science is a new arithmetic because that's a really good way to look at it. It's kind of common sense. What's your advice? Great question. I have kids of my own. But, you know, that talking about kids requires a beer. So I have to have one here, luckily. Cheers. So, you know, but no, education's got to change. You know, we still teach long division to kids and I don't really think that's the priority. Yes, it's nice to learn algorithmic thinking, but it's really better to learn about uncertainty and how was that data collected? How do you make an inference? I mean, all human beings collect data around them about their friends, about the world around them. And then they're still uncertain, but they have to make decisions. Which school to go to? Which friends to have? What to do on the weekend? And we're good at it, but we evolve in a world where there wasn't that much data. There was just what our eyes could see and what we could behold. And now all of us are surrounded by all kinds of other data that's sampled in complicated ways with people adversaries and people want to sell you stuff. How do you filter it out and still make good decisions? And so, yes, learning in school about not long division but learning about Bayes theorem or learning about the bootstrap or error bars should be part of everyone's education from the get go from age five up to through college. Any advice for the parents out there that might be watching about kids? Let them be free, open their brains up. It sounds libertarian, but in a way, this is the new normal. Well, yeah, it's, where's my beer? Yeah, it's about being an individual. We're not having to all watch the same channel anymore. You know that you're not Walter Cronkite. You're just some guy with a microphone, right? And the people who are interested in you follow you and that's liberating. And we're all interested in certain kinds of music. We can get it. We can get recommended for books, for instance. So it's a liberating world to be in if you don't get too confused by it. But embrace it. You know, I wouldn't be, I'm not an afraid parent. I think it's all liberating and I would let your kids teach you and follow them. Well, certainly theCUBE is a big following, as you know, we've been, you know, on the ground. And this is social media and social media brings this new generation question. So I have to ask you around computer science and topic near and dear, my art, I think a CS degree. What is the future of the computer science curriculum? Obviously you bring up some really touching things about societal benefits. There's a lot of things going on in our society, not just in the North America, but globally, global economy, connected devices, cloud, integrated stacks, whole new paradigms shift, whole new worlds coming. What's your take on the computer science next step? What's your vision? Well, so what's great about computer science, I think is the focus on abstraction. Think was on modularity, divide and conquer, building pieces. But what's been missing critically is the notion of uncertainty. And so, in fact, I looked recently at a proposal for the curriculum for the next 10 years. And, you know, many current themes were in there, but there was no notion that an uncertainty should be infused throughout all the cycle of taking in, of processing and algorithmic thinking. So computational thinking needs to come together with statistical thinking. That's the future of both fields. Michael, thanks so much for the time. I want to ask you one final question. What's the coolest thing you've seen over the past year around tech, science, life, society? Cool, oh boy, you've got me there. I'm a musician and I love the fact that I can get on YouTube and I can see how some drummer actually played that song. And just the, that's not cool, that's not high tech and nowadays, but it's just so liberating to be able to sort of have that kind of thing at my tip and fingers, you know? And that's a big change to all of us. Great, new user experiences, new opportunities, new opportunities for coders and for businesses. This is The Cube, I'm John Furrier. Thanks for watching.