[MUSIC] One very important analytical technique in computational social science that we will surely have to talk about is basically artificial intelligence. So we will talk about machine learning, and about natural language processing. The idea behind machine learning is basically that we say, well, we have so much data nowadays, the data is so big, big data that this information processor here is helplessly overwhelmed with processing it. So, the reasonable thing to do is we got to fight fire with fire. So the digital technology broad as this information overload, let's also use digital technology in order to make sense of it, all right? So if we use machines to learn patterns in this information overload. So for example, think about books. If you read one book per day, for your entire life, every day you read one book for 80 years, you will read about 30,000 books. Well, that's a lot, and hopefully you'll be able to process most of it in this processor here. Now Google books has 130 million books, like there's no way shape or form you can process that. But Google uploaded them, so they are all there. And actually, with machine learning, we can look for patterns. We can look for what's there and what kind of patterns and new patterns maybe that we rediscover of taking, looking at 130 million books, not only is about 30,000 that's insane to begin with. But we can look at, well, the patterns across many different data inputs and machines can do that. All right, so machines looks for pattern, discovers some of these benefits, and then what can they do with it? Well, they can do a lot with it. Actually modern approaches to artificial intelligence only became viable with this big data footprint. The idea of artificial intelligence has been around for a long time, but things to this digital footprint that we have, and thanks to the massive amount of data now, machines can learn discover patterns and do really amazing things. So check out some of the videos I have here for you. This for example up here is the Amazon inventory warehouse. So if you click something on an online retailer like Amazon and you want to buy something, immediately, one of these bots, artificial intelligence bot gets going and gets your purchase This over here in the corner, that's what a modern car manufacturing company looks like what it is, there's no person involved here anymore. It's machines who learn the patterns of how to construct a car. It's hundreds of robots working together, and if you find a human, it's probably some kind of IT nerd, who adjust the robots but cars nowadays are basically built by artificial intelligence. And this down here is an example that I want to show you, is how quickly these machines can learn. So this is an example of an artificial intelligence from Google that plays a very old Atari game, I used to play that as a kid. And the artificial intelligence doesn't know anything about the game. It just knows, it's kind of like a ping pong ball, and it has to collect points. So that's the only two things, you tell it, it t doesn't know how the game works. And that's how quickly it learns how to master the game, check it out. Wow, that's pretty impressive, right? Algorithm became innovative. It actually innovated, it figured out that putting the ball on top rattles everything. It found a solution, a shortcut. I played this game as a kid, and probably for more than a few hours, and I didn't figure that one out. I mean, there are people who figured that out, but honestly AI want me for sure. I mean found a very creative solution for that, and that's what we often find. So if we led machine learning lose on this data, it find some kind of patterns that this process here was not even aware of that existed. Sometimes also be looked at this black box of machine learning and we don't even still don't know after looking at it what it actually did, and why it is so much better than humans in doing what it does. And it became better than humans in many, many areas, for example in radiology, the discovery of cancer cells. Nowadays, artificial intelligence there is more reliable, and of course with self driving cars and we could go on. It's more secure even, then people we can go on and on artificial intelligence is winning us every day with something. And it does so by looking through these massive amounts of data and discovering patterns, creating knowledge, and that's what computational science is about, right? So machine learning can help us with that.
