History of the Singularity University and Artificial Intelligence

Okay, well, as Erik said, on a day-to-day basis I'm the host of a radio program called HotDisc at the Danish Broadcast Corporation, but I'm also a former pupil or an alumni of something called Singularity University in Silicon Valley. Erik briefly mentioned them. We have something called the Singularity U in Copenhagen, which is an association of old alumni from Singularity University. I'm not going to go deeper into that, but you can always Google it and find more about us on the net. I promise to say a few things about Singularity University and then dive deeper into artificial intelligence. This guy is one of the founders of Singularity University. He's called Ray Kurzweil. He was mentioned before. He's an inventor. He's a writer. He's a scientist. He's also the director of engineering at Google, and he's been there for like three years now, working on chatbots and artificial intelligence and stuff like that. And he's sort of a controversial person in many ways, but we will dig deeper into that later. He co-founded Singularity University together with a businessman called Peter Diamandis. He was also mentioned in Erik's talk. And they did it together with a bunch of people from Silicon Valley who all came together back in 2009 to work on the idea of the Internet. So it's a very interesting idea about exponential technology and how it will affect the world and how we can build businesses to do better for the world using this exponential technology. The thinking about the exponential technology at Singularity University is in part or mostly based on the thinking of Ray Kurzweil. And he has this idea that technology evolves according to Moore's law, which will take us to a point in time where we will have computers that are better at calculating. I won't say intelligence, but they have more calculating power than the whole human race together. And that will be available for $1,000. So we have this exponential curve. And when we reach that point, he has predicted that that would happen in 2045. But already in 2029, we will have really powerful computers that you can compare to human level intelligence, or at least calculating power. So it's quite near in the future. This point in the future in 2045, Ray Kurzweil talks about it as Singularity. Singularity is sort of a black hole in the universe where all the information is very condensed, meaning that you cannot really grasp what will happen outside the black hole. Outside the similarity. You cannot see beyond what you would call the event horizon. And that sounds a bit severe, a bit science fiction-like, but we have actually had a lot of these similarities through human history. I mean, you could, for instance, think about when we climbed down from the trees and we started walking in two legs, and we walked across the savannah in Africa, and we invented fire, and we invented language, and so on. Later on, we moved together in cities, and we invented writing, and we invented printing press, and so on. Personal computer. What you see is that these inventions, they come closer and closer, which means that the speed is increasing. And what Ray Kurzweil claims is that in 2045, due to exponential technology and computers and so on, the speed will be very, very high, and we will have this total, singularity. And that's probably what Nick Bostrom and the others in Elon Musk and so on are very worried about. Somehow, it's a bit like what Marcel McLuhan talked about back in the 60s, that technology will become the extension of man. Ray Kurzweil, he goes a bit further. He sees us as merging with the technology, so we sort of converge with technology, and we become technology. Part technology, part biology. Um... This is some of the thinking that you are being taught at the Singularity University. But it's not all about artificial intelligence. They also teach a lot about other technologies. Artificial intelligence is one of them. Nanotechnology, synthetic biology, genetics, robots, 3D printing, and a lot of other stuff. But if you remember the watch list from Nick Bostrom, the things you should look out for, the dangerous stuff, I mean, you could count some of them in there. Synthetic biology, for instance, genetics, personal genetics, and so on. But it's all about understanding that technology evolves in an exponential fashion. You can get it, it becomes better, I mean, your phone becomes better every year, but it comes at the same price with me, which means that you actually have a very cheap supercomputer in your pocket, and you're connected to the rest of the world with this one. Um... At the Singularity University, they're not trying to build machines that will destroy us. They're actually all killers or anything. They're actually trying to, as it says here, educate, inspire, and empower leaders and other people and leaders as well, also startups and students, to apply exponential technologies to address humanity's grand challenges. And what are these challenges? Well, they are gathered in these nine points. These are the problems, problem fields, I should say, that Singularity University is using exponential technology to solve. And the idea is, of course, that you go there and you learn about these exponential technologies and you leave, and you do a start-up, trying to solve some of the greatest challenges that humanity has. So it's not about building AI that will take over the world. It's actually about solving a lot of the very serious problems that we face as a human race and as a planet. They have something called the Global Impact Competition. You can actually apply to go there for a ten-week summer course. Every summer, they... take in... That's not the right way to say it in English, but anyway, if you apply, 80 people will be able to go there. And some of them won't. Maybe you don't have the money to do it, or you need some help to sort of fine-tune the idea that you will present to them. And then you can participate in this Global Impact Competition, where you submit your suggestion and the curriculum, and then you will be considered as a candidate for being there for ten weeks. And there's a small video where I care to correspond. It actually explains what the purpose of this is. The purpose of the Global Impact Competition is to create a Silicon Valley culture and reality in every country. Everywhere I go in the world, people tell me, oh, our neighborhood is the Silicon Valley of France, and we're the Silicon Valley of Israel. What is Silicon Valley? It's a metaphor. The ability of individuals and small groups to transform major industries and major ideas to make the world a better place. Hi, my name is Susan Graham. I'm from Sydney, Australia. I'm Hussam Zirkelsen from Denmark. My name is Cosma. I'm from England. I am Leo Valente. I'm from Argentina. And the GIC project I was working on is planting a billion trees to reforest areas where currently we're using linear techniques. I had the idea of making a collaborative tool where people that know how to do CPR could come to rescue someone in the near perimeter who has a heart attack. My project used patented image processing technology to monitor the healing and progression of chronic healings. My project is a platform that enables small businesses in developing countries to accept payments with the so-called social cards. As a GIC sponsor, you are not just enabling one person to participate in the grant studies program, but you will catalyze the community's enterprise. by providing a solution to the needs of the population. In Denmark, we had to impact 10% of the population. Even though it's a small population, Denmark is a small country, we had to come up with a solution that would impact them already within five years. So it had to be very concrete, very practical. Social entrepreneurship is something I've become quite passionate about. And the GIC embodies this. If you want to be a GIC sponsor, a contestant, or want to learn more about the graduate studies program, please click on one of the links provided. We run a GIC Global Impact Competition in Denmark every year. It's called Danske Ideen, and it's part of the Singularity UKO main chapter. But I actually talked about this yesterday in this very room to a significantly lower, significantly fewer people. But now you have heard about it also. And then we will have to dig deeper into intelligent machines. So I was thinking to sort of have another angle on this, which is not the scaremonger, but a much more angle on artificial intelligence. I was thinking about what are we really using these machines for? What are we using artificial intelligence for? How do we actually communicate with it? And what do we want to do with it? Erik already talked about how we are being beaten by... This should actually move. It doesn't move. Now it moves. Hey! We are constantly being beaten by machines. And it seems to be our fate to be beaten by different machines, like chess or Jeopardy or whatever it is, or maybe it's Go. And I was looking at this guy, Lisa Dahl, who is the world champion in Go. And as you know, he was recently beaten by Algebra, which is an artificial intelligence or a recurrent neural network created by DeepMind in London. When he was beaten, it was transmitted live on YouTube. He had 100 million people looking at this stream. Six million of them living in China. So it's really huge. And it's like, if you could imagine the Danish soccer team being beaten by a bunch of robots. I think that's the way they felt about this. So he was really... He was not in a good mood when he had gone through this. And he was sorry. And I was thinking, first of all, I was thinking, okay, why is he sorry? Who is he sorry for? Probably he's sorry for the 100 million looking at him being beaten by this artificial intelligence. But maybe he also apologized to the whole of humanity, to all the people in this room. Or maybe he actually apologized to the machine for being such an awesome player. I'm not quite sure. But this guy, Jorm Smyth-Huber, not a lot of people have heard about this guy. He's actually a pioneer. He invented or started making recurrent neural networks in the 90s. And these are some of the networks that are behind stuff like the DeepMind and other machines that we would call artificial intelligence. I met him six months ago. And he told me, well, artificial intelligence, they do not have a conflict with humans. And if you kind of read what's on the bottom, it says super smart AIs of the future will be mostly interested in other super smart AIs of the future. Not so much in frogs, not so much in ants, and not so much in humans. So what he's saying is that if there is a conflict in the future, it will be a conflict between AIs and not a conflict between AIs and humans. That's his claim. He has made character recognition, recognition of natural languages. He's responsible for an ad where you can go to a Chinese restaurant and use your smartphone to read the menu, although it is in Chinese. He has been responsible for making self-driving cars better, recognizing street lights, and stuff like that. And he's done it in his research lab at the University of Lugano in Switzerland. And actually, his algorithms are now also used by Google and others. So he's really sort of the godfather of recurrent neural networks. And why do I bring this guy in connection with the AlphaGo? It's because 20 years ago, I believe, I read an article or an interview with Björk, you know, the Icelandic singer. And she talked about how she started to use the computer for making music instead of having a band. And why did she do that? She did it because she was sort of having a conversation with herself. The computer was a mirror of herself and her thoughts. So she was sort of like thinking with herself, communicating with herself. Which means that what I'm thinking here is that an artificial intelligence might not be something alien to us. It might just be a reflection of ourselves. It might just be us talking to ourselves the way we program it. Which brings me back to Ray Kurzweil and the Singularity University. He has mentioned that when you talk to a human in 2035, you'll be talking to someone that's a combination of biological and non-biological intelligence. Which is some of the things that he's constantly saying and it freaks people out because they're sort of like, what are you trying to build? And he has been working on creating his own avatar, which by the way is female because that's more interesting if you become a female in virtual reality. You might discover some new aspects of yourself. But he has also been working on a very controversial project, which is bringing his deceased dad back from the dead. So here's a video where he explains it. Do you think you can bring that to your father in a way that he would have a continuity of consciousness? I've got hundreds of boxes of documents, recordings, movies, photographs, I've made a process of digitizing all that. And actually a very good way to express all this documentation that my father, who was a musician, and who wrote his music, and who wrote his letters, and movies and other artifacts, would be to create an avatar that an AI would create. That would be as much like my father as possible given the information we had about him. And including possibly his DNA. And that person would be very much like my father. You could argue that he would be more like my father than my father would be had he been alive now, which would be close to 100, so it's possible he could be alive. But he wouldn't be very similar to the way he was when he was 58. Or 40 years ago. And he would pass a Frederick Kurzweil Turing test with me as the judge, or anyone else in the union. And the Turing test, named after the human body, now, Turing, what is that? The classical Turing test is you have a human judge, you have an artificial intelligence, and a human foil, and several of each. And the human judge interviews the both. Over all the time. Over all of the events. You can't see if she can't tell the difference between a humanist who passes the Turing test. And in my view, if an entity passes the Turing test for being human, let alone a specific person, that person is conscious. Now you raise another question, okay? This avatar of your father being a conscious person, and it admits you that it's your father, and it's not your... It remembers of him. But that doesn't mean it's actually his consciousness. It doesn't have to be that way. That thing is really... These are philosophical issues. Which means you can't really decide on those. So... How many of you think this is a good idea? Very few. Someone is in doubt. I think maybe we should discuss it afterwards. Funny thing is, five years ago, Ray Kurzweil was in Denmark. He was going to do a talk in Odense at a conference for blind people. He's done a lot of technology there. Helped blind people, actually. And his flight was delayed. I was the moderator for his talk. His flight was delayed, and he was supposed to arrive the night before. And then he arrived just in time to make the talk, and he was totally jet-lagged, and his suitcase was still in New York. It was a big mess. And he had to go to London directly afterwards. And I was promised to have an interview with him, but he didn't have time, so I jumped on the cab with him to the airport, from Odense to the airport. And first of all, he was sort of confused because his suitcase was gone, and he had to call the secretary back in the world, but he's Blackberry. I couldn't believe that this guy, who's sort of the father of the country's intelligence, he had a Blackberry. But it didn't work in Denmark, so I gave him my iPhone, and he didn't know how my iPhone worked. I mean, apparently he had never seen an iPhone before, or seen one. So it was a bit of a confusing drive, but we managed to get hold of his secretary and talk about his luggage, which would arrive in Copenhagen a couple of hours later than he had left Copenhagen, which was sort of a problem. And then I wanted to do the interview, and then he fell asleep. And he was like hanging like this from the seat in the car. And I was sitting there with him, sleeping, and I wanted to do this interview with this guy, and I wanted to ask him, about his father and his official intelligence and so on. And I was sort of like, I mean, can I wake him up? What do I do? But apparently, suddenly he woke up and he said, well, let's do the interview. And we did the interview, and we talked about his father, and we talked about what he was explaining in the video. And first of all, my feeling of this guy before I met him, and sat with him there, I mean, he was so human, just sitting there sleeping, he was tired and so on. My feeling before that was that he was sort of a freaked out scientist who wanted to create this copy of his father. But then we just started talking about what he actually wanted to do. I mean, his father was a musician, and he wanted to play piano together with his father. So what he wants to do is to create this sort of artificial intelligence, this copy of his father, which is just as good as it's, that it will fool him to believe that it is actually his father. And then he wants to play the piano together with him, which is very human, and not very crazy scientist-like. So actually, when he left the camp, I had this feeling that I had met a small boy who was actually just longing for his father to be alive and play piano with him. Then the other day I started thinking, well, actually, what he was talking about here was something that we didn't really have at that time, namely a chatbot. Today, chatbots are immensely popular, and we see them popping up all over the place, and they're basically immediately on neural networks that we can chat with online. We had a very... ...a very unsuccessful chatbot called Tay, which was created by Microsoft like two months ago, I think. It was supposed to be a teenager. It was supposed to learn from other people on Twitter. And it went horribly wrong, because in like 24 hours, it became a total... ...racist. So it had to leave Twitter. And it wasn't supposed to come back. And it reminded me of an interview I did six years ago with the guy who holds the chair of artificial intelligence at the Singularity University, a guy called Neil Jacobstein. And he told me that at that point, because we talked about... I mean, if you want to... Boom, boom. There it is. Am I having problems with the battery, or...? Okay. You can still hear me. We talked about the ethical dimension of artificial intelligence. How will you put human values, ethical values, how will you program that into an artificial intelligence? And Neil, he wasn't so worried about that. He said, well, we can raise our children. They're savages when they're born. I'm not sure whether he has children himself. But we can civilize them, and we can do the same thing with machines. And when I saw this Che thing, I was like, okay, so we can civilize this chatbot on Twitter. I'm not quite sure about that, because they took it offline, and like two days later, it came back. And it had been smoking pot in front of the police. So what happens when you put a teenager into house arrest? A teenager... Well, the teenager escapes and does something that is even more horrible. Anyway, this is mostly your... funny. But, I mean, we are not quite there with this artificial intelligence. We are, apparently, some years from 2045. But maybe we're just building something which is immensely more boring than having your father as a chatbot or building an artificial intelligence which is capable of running the world. Maybe we're just building sort of invisible servants, invisible servants that we can call up on our mobile phones, in about a few years, in 2020, 80% of all adults on the globe will have a supercomputer in their pocket, a mobile or smartphone, and you can actually access artificial intelligences or chatbots or neural networks through this iPhone. You can do it already now. And as the guy, Jorg Schmidt-Huber, I talked about before, he has said, AIs that we have access to are all designed to make us even more addicted to using the phone. So maybe that's the dangerous part about the AIs. Another thing we should take into account is the MakeCom's law. That's the law about the network effects which says that the value of a network doubles exponentially as you add users. And some of the users on this network are not only people, there are also a lot of different devices. We will have 50 billion different devices on the Internet, that's what is called the Internet of Things, by 2020, which means that we will actually be communicating with a lot of machines. So what are we doing when we communicate with these machines? We are actually teaching them, like we were supposed to teach Tay to be at least a teenager, we are also teaching these machines to recognize pictures of cats, dogs and so on. And there are a number of them that have sort of converged and come to existence during the later years that has made this possible. Neural networks, we have known them since the 80s. The problem with the neural networks that Jorg Schmidt-Huber invented back in the 80s and the 90s was that it didn't have the computing power and it didn't have sufficient data to train them. Well now we have the GPUs, the graphic processor units from the gaming machines which are really capable of running high speed calculations. So they are used in artificial intelligence. We have a lot of big data which is mainly made by us using these devices. A lot of pictures on Google and Facebook and so on and video. And then we have the network effect that I just discussed, that I just talked about. So all this being interlinked. So we have a lot of things, a lot of technologies coming together creating this space of artificial intelligence making it possible. Now some people don't think, don't believe in the singularity. Some of you may know this guy. He is called Kevin Kelly. He is an author and he is a thinker from Silicon Valley. And he was also one of the co-founders of the magazine Wired back in the day like 25 years ago. He does not believe in the singularity. He does not believe that we would have this giant artificial intelligence that would sort of change the world in a very severe way. He actually prefers to talk about it as being artificial smartness rather than artificial intelligence. And what he is saying is that all the gadgets and all the stuff we have it will have an added layer of smartness on it. And that is actually how it will be. One of his examples is the self-driving car. Of course the human driver will disappear. But the robot drivers they will not be anything like human drivers. Because one of the things about self-driving cars is of course that they will decrease the number of accidents because they will not be drunk. They will not fiddle with the radio or they will not text on their phone and they won't talk to the children in the back. They will actually only be driving. They won't be distracted. So you will actually develop something that has no conscience. And it is pretty important it doesn't have any conscience Which is sort of boring if you think about AI as something really glorious. The other thing he says then is that the AIs of the future they will live in the cloud as a service and it will become a commodity and it will be like electricity. And I mean who can get excited about electricity? I mean it is just something that we have all the time all day and night. And artificial intelligence will be like this if we should believe Kevin Kelly. It is going to be pretty boring and it is going to be nothing like this guy. And as Jorg Smithofer puts it these things that you see now in Schwarzenegger movies are very unrealistic because there is always some super robot and he seems to have conflict with humans which doesn't make any sense at all because it is just going to be something pretty boring and dull happening in the background in our daily lives. Yeah. And I believe the questions will be after we have the last talk. And he is already with us on Skype.