[MUSIC] Maybe we're passing the top of the AI hype curve at this moment, but maybe now it is a good reason to really care about it. [MUSIC] One might wonder, how come we talk so much about AI, artificial intelligence, at this moment? What's actually happened out there? And asking that kind of question every time we find a new technology popping up is most often useful, because we will find some learning into it. And going back in history, AI is actually not new. Research has been ongoing in the field of AI since, roughly speaking, the 1950s. So why do we talk about it in practice nowadays? [MUSIC] >> Deep Blue versus Kasparov, Chinook versus Marion Tinsley, and now AlphaGo versus Lee Sedol. Lee Sedol walked into the conference room, receiving a massive applause and more congratulations than any game he has ever won in his life before. >> [APPLAUSE] >> The suddenly bleak and gloomy clouds clouding the Go world have dissipated, and everyone was cheering twice as hard for Lee Sedal in game 5. Well, ultimately he lost by two and a half points, but this is the closest result we have ever seen thus far. AlphaGo is suddenly not the insurmountable wall that we thought it was. This proves that Go is still quite alive and humanity is not quite done just yet. [MUSIC] Well, for now anyway. >> I'll be back. >> The new version of the Atlas Robot by Boston Dynamics is the most advanced human walking robot ever invented. It can not only stand and walk by itself, but is seemingly quiet and has an advanced sensor system that means it can walk on almost any terrain without falling over. This is a major breakthrough. It uses sensors in its body and legs to balance, and it uses stereo sensors in it's head to avoid obstacles, assess the terrain, help with navigation, and manipulate objects. What do you guys think? Aren't robots advancing at an amazing rate? How long will it be until they are a part of our everyday life? [MUSIC] >> Well, roughly speaking, the algorithms used for AI, the mathematics [SOUND] is really not the new thing that's happened during the last decades, it's not the main reason for talking about AI. There's actually two reasons. [COUGH] One is computer power [SOUND] or roughly speaking, Moore's law has given us reason to care about AI at this moment. The thing is, what we can do with AI publicly today is something we always have been able to do since the 1950s, but it was extremely complicated and extremely expensive. Now it's rather cheap to do it, and that's why we get it into the public sphere. And the reason why it's become cheap is because of Moore's Law. Processing capacity, storage capacity, the cost for it has gone down and capacity has gone up. So now it's economically possible to do what researchers in their laboratory were able to do during the 1950s, and that is a big difference. Then there is one piece more, and that's the availability of data, because we need that kind of data to build something with AI. And with the Internet, we've increased the amount of data existing, like for instance, the amount of pictures, the amount of movies, the amount of songs, the amount of websites. The amount of texts available on the Internet nowadays, making it possible for us to build AI, and now we have it in the public sphere. So in that sense, it's Moore's Law and data, that is the reason why we start talking about AI now suddenly. Well, actually not suddenly, looking at in hindsight, we could, roughly speaking, have estimated only by understanding Moore's Law and its development. That AI would be something that we would be talking about at this moment. Now, having that said, algorithms are not that really complicated and unique, at least not unique. Nearly everyone can have access to them. Computer power is also something that everyone can have access to it. But data is something that not necessarily everyone can have access to. I have data, my own data, I'm the only one having access to it, then it's a unique kind of asset. That's what's happening gradually in the field of AI nowadays, that assets are changing character, or roughly speaking, data is becoming a valuable asset. So the battle between the big actors when it comes to AI could roughly speaking be called the battle among the ones having the best kind of of data, and different actors get different kind of of data. For instance, what kind of of data do Facebook have? They have data about who we are, but not necessarily the kind of of data about what we want, that might be something that actually Google is having, as a search result. But on the other hand, Google does not necessarily know who we are, who is searching for something. And then for instance we have Amazon, what kind of data do they have? They have data on what we do purchase, but not necessarily data on who we are. Looking at it in that way, we could find possible new kind of interesting players in this field. What kind of data do different kind of companies already have, so to speak? And what kind of activity do they have that gives them access to certain kind of data? And what kind of players to actually already have a lot of data? Well, one of them are, for instance, telecom operators. Another one is energy suppliers. So looking at it from a data knowledge, data source point of view, could add knowledge into what will happen with AI in the future and not the least why it has happened already. Well then, how is the discussion about AI at this moment? Well, if we take a look on the Gartner hype curve, [SOUND] we might say that we're on the high top at this moment. The hype curve is actually only measuring how we talk about new kind of phenomena, like new kind of technology. Ten years ago, we didn't talk that much about AI, now we talk tremendously a lot about it. And they measure, with the Gartner hype curve, the interest in certain kind of fields. So it comes, goes up, and then after a while, the interest goes down, mainly because we overestimate, at least in the public media sphere, what the new phenomena can be used for, and then it becomes a backlash. Everyone starts saying it didn't happen, it wasn't that good. But then normally that is when it does happen. So maybe we're passing the top of the AI hype curve at this moment, but maybe now it is a good reason to really care about it. And what that nearly tells us every time a new phenomena pops up is that in the short run we have a tendency of overestimating what it can be done with what we can do with it. The Internet is a really good example of that. In the early days of the Internet we had huge visions that never came about, but is about nowadays, 25 years later. So we overestimate what we can do with it in the early phase, the short run, but in the long run we have a tendency of underestimating the consequences of it. So having that said, I'm not sure if we really should care about what media is saying about AI. We should use our own need and our own interest in the technology to decide whether we should care and how much we should care. But on a society level, it feels rather reasonable to assume that there will be backlash when we talk about AI, but then it might be, when it really does happen. [MUSIC]
