 Hi everybody, we're back. This is Dave Vellante. I'm from wikibond.org and this is Silicon Angles theCUBE, where we bring you the smartest people that we can find. We extract the signal from the noise. We're here at a Riley media strata conference. It's day two for us. This is the, I guess the fourth time we've done strata, maybe the fifth, but this is the third strata in Santa Clara. We've been at all three, and one of the first strata's we ever did was an amazing event, Silicon Angles. Cube got huge, huge traffic. You were all very, very interested in this event. Big data was just coming in to the discussion in that parlance. And Joseph Turian was one of our original guests at that event. And Joseph is a PhD data scientist, machine learning expert, AI expert, entrepreneur, and probably 100 other things that we'll find out in this interview. But Joseph, first of all, welcome back to theCUBE. Oh, it's nice to see you again. Thank you. A lot going on, as I described. You were talking off camera. You got some new project that you're very excited about. I want to hear about that. But tell me, you know, what's new in the world, in your world, what you're seeing that's interesting, challenging, and things that are attracting your attention? Sure. So I mean, actually, the last time I spoke with you, we were talking about deep learning, which was this wacky new thing in machine learning. It was kind of an academic thing. But now we've actually seen this year it's ready for prime time. We're seeing deep learning being used in industry, right? Google came out, they had a very public paper where they actually watched YouTube videos using deep learning, and it started just picking out cats without being trained on anything. That was kind of a fun paper, right? But more interestingly, they're using deep learning in speech recognition in Google Now. They're trying to put deep learning into a lot of their products as far as they've told us. Microsoft is also using deep learning stuff. They had a demo where they showed someone speaking, and it took his speech, translated it into Chinese, and then spoke it as if he were speaking Chinese. And they're using deep learning for this. So a lot of the stuff in machine learning that academics have been working on is now actually we're seeing it in production, in an exciting way, for things like visual recognition, speech recognition, translation. All the stuff people have been talking about for AI for a while and wanted, we're actually starting to see. And I also believe they might be using it for their self-driving cars. So this is something that's I think an interesting job. So really challenging problems that are starting to find their way into potentially commercial applications with much greater degrees of accuracy and utility than we've seen in the past. Yeah, and I mean that was the thing that was really excited about deep learning was this idea that it was ambitious, but also very rigorous. So it wasn't just a lot of hype, it wasn't just a lot of talk about, oh we can do it like this and this looks like the brain. No, these techniques actually work and we can put them into use. And that's what's really exciting is seeing these ambitious projects like self-driving cars and speech recognition, translation. Really actually perhaps in our hands pretty soon. So Joseph, I mean artificial intelligence obviously an area that you know a lot about. It's been around for a long, long time. It's had a, it's gone through cycles of ad-raps and obviously making its way back. What has been the enabler? I mean obviously data is one of those enablers, but what are the things that have catapulted deep learning, machine learning, artificial intelligence up back to sort of prominence and great potential? Yeah and you're right that we've been burned several times by hype around artificial intelligence and people have been very leery of even the term artificial intelligence. If you were a serious scientist you would not say artificial intelligence because you'd be like, people would be, no, no, no, you must be a coo, you know? So first of all like there's this leariness of artificial intelligence. But you've embraced it. That's what I love about it. No, I mean it's finally getting to the point now that we can start saying again it's not dirty word. But serious scientists, you know, for a while being burned and saying, you know what, first there was this camp, it was very popular in the 90s. They said, we're only going to work on problems that we really understand what's going on. And then a sort of a faction of people said, you know what, we actually want to, we want to really push the envelope. We're going to try some new stuff. It doesn't matter if we don't exactly understand what's going on because if it works and we can demonstrate empirically that it works then that's exciting in itself. So these guys who made this decision, we're going to work on things, even if we don't completely understand why it works, if it's effective and rigorous, then we will still study it. So it was kind of a change in thought. There was one of the things that led to this, these advances, a sort of new sort of ambitiousness while at the same time staying grounded and realistic, you know? So you talked about some examples that Google and Microsoft are doing with video in particular, which has been a really challenging problem. What other applications do you see that have great potential that are exciting you? Well, so I know if you follow Kaggle, Kaggle did a competition for Merck on drug discovery. And the people that won the drug discovery competition, if I recall correctly, they used deep learning. So all the things right now that people are doing machine learning for, whether it's financial modeling or medical applications, all these things, if you're already using machine learning, perhaps we can use deep learning and push the envelope even further. That's the idea. You know, I asked you a couple years ago about Watson. Watson was just coming out and you were like not too excited about it, right? Okay, so. Okay, but of course at the time it was new and you know, the great demos. I mean like, you know, supercomputers playing chess at the time were pretty cool. But from a scientist perspective, it was like, okay, you know, that's nice. But it was, I think you, my words, you've described it more of sort of a brute force approach. So can machine learning, you know, provide a more elegant or efficient or more powerful model than say for example, what we've seen with Watson, even though it gets a lot of buzz and a lot of attention? Well, I don't know. I mean, Watson actually is a pretty good case study because you know, the world is a messy place. And if you want to build something real, there's going to be a lot of different parts that you have to glue together. You know, and Watson is kind of a Frankenstein and it's not that elegant, but it does work. You know, and that's actually kind of great. So I mean, machine learning in the real world is probably going to be a piece of the puzzle and this puzzle is going to be pretty big. Okay, so it doesn't have to be all shiny and polished and pretty. I mean, that's not how anything in the real world is, right? So you were telling me about this new project without giving me any details. But what can you tell me about it? I can't talk about it right now, but it will help you, it will help everyone understand their data in a much more easy and accessible way. That's it. Here, let me tell you what I think is cool right now, okay? This is something different. It's, so I'm really interested in right now in how, you know, Mark Andreessen says, software is eating the world. And we're seeing, I think really cool examples of this. There's collaborative consumption, things like Uber and Lyft and Sidecar where people are really solving problems in the real world and logistics using software. I think that's one thing that's really exciting. And then coming at it from another direction is this whole internet of things where there are sensors everywhere and sort of in the middle is also like Google Glass, right? Which is like an embodiment of software that we can interact with. So I think that's the other trend that I think is really exciting right now is just software helping us in our physical reality and becoming much more pervasive physically. So there's notion of wearable devices. It was really quite interesting, particularly from the user experience standpoint. I mean, it completely changes the, I don't even want to call it an interface. So I'm not sure it's an interface anymore. Yeah, wearable devices. And then, but then there's also stuff like, you know, I think Uber and Lyft and Sidecar, I think this is very exciting, right? And then there are these new services like eBay Now. You press the button on your phone and in two or three hours they get you what you wanted, right? Or Postmates, they're doing the same thing. It just errands on demand, exec is on demand, right? This idea of just every single thing that might be physical being optimized and much more efficient using software I think is a really exciting. So what's most interesting about that from a consumer standpoint is it simplifies things. Hadoop is still really hard, right? I mean, the number of people that can actually do something with it is quite limited. So will those two worlds meet? I don't know. I mean, eventually they will, how quickly they will? I don't know. Is that something that you'd like to try to accelerate? I mean, you wouldn't tell us anything about what you're doing, but is it around sort of simplifying the complex? Is that a fair bet? I mean, so, you know, I think that accessibility is where everything is moving. And accessibility and simpler interfaces is the real theme that I think we're getting at, right? Because Google Glass is just a much better interface than playing around with your thumbs, right? Using your voice is a much more accessible interface. Pressing a button on your phone is much simpler than going outside and hailing a tab. So I think the move is more towards accessibility and towards naturalness of interaction. I guess that's the thing. So watch, talk about Apple doing a new wrist device, like the Tracy or 007. Yeah, I mean, I don't know. I think just speaking your voice and not having to even lift up a device is even easier. But it seems like there's this push towards just everything becoming more accessible, you know, and I think that's exciting. How about, you mentioned the internet of things. And I think of the security problems around that. I think of what Intel announced this week and how they're trying to attack that problem. I think of Stuxnet I read earlier this week that actually was developed probably around 2005, which doesn't surprise me, but the most conventional wisdom we found out about it in 2010, I think. So what are your thoughts on how that whole internet of things will emerge, how the security model will have to change, and the simplicity model, the interface to those devices? No, I mean, security is like a huge question. I don't even know if we can get into that, right? That's like talking about what's going to happen with privacy in 10 years. This is such a huge question. Well, let me ask you. And it could go so many ways. Let me be more specific. Do you think that security is a complete do-over in order to be able to accommodate that type of infrastructure? I don't know. That's a really hard, it's just an impossible question to ask, you know? I'm going to have to give you no comment on that one. No comment, all right. All right, fair enough. What else do you want to talk about? I don't know. So I have a startup idea for anyone that wants to do a start. So here's a good startup idea. So we were talking about collaborative consumption. Here's the idea. A lot of people have stuff in their apartment that they don't need and they want to get rid of. But they don't want to go on Craigslist and Celepics because that's a pain. Other people need stuff, but they only need temporarily. So the idea is Uber meets consignment, okay? Or Uber meets storage. We come, we get your stuff, we take it away. If we can sell it, then we will. Otherwise, we store it and you pay us on a monthly basis and compress the body and get it back when you need it, right? Because people talk about like, oh, what'd it be great if I had a co-op where I could share hardware like for home repairs and stuff like that? But what I'm proposing is even easier. You buy the hammer, then when you're done with it, you give it back and if we can sell it to the next guy, then you finally cash out. That's the idea. If you want to do something cool, I want this application right now. There's too much stuff in my apartment and I want to sell it, but I don't have the time. It's kind of a modern version of an auction house. Yeah, exactly. It's Uber meets an auction house. Yeah. That's good. All right, Joseph. Well, listen, I really appreciate you sharing some ideas, respecting you, not wanting to talk about security because it's such a hairball. It's a can of worms. Bowl of worms, bowl of ox. And Joseph says no comment on security be a do-over. I actually think it pretty much is a do-over, but we'll see. You might be right. And I hope so because every year I look back and say it's further and further away from actually succeeding, but it's a hard problem. But anyway, it's always great to have you on. We'll be tracking. Please keep in touch. Let us know. What's the timing for when you can actually talk about your new venture? When it's ready. Within the next year? Within the next few years? Definitely within the next decade. Within the next decade? Yeah, within the next decade. Okay, good. So that's a long runway for innovation. So we've made more of it. Hey, change of the world is hard business. Joseph, great to see you, man. Thanks very much. All right, keep it right there and we'll be right back with the show wrap up here live from the Stratoconf in Santa Clara, California. This is Dave Vellante of Wikibon. This is theCUBE.