 First of all, thank you all for being here, and I'm very glad to be back and I feel like an old customer in an old pub, even though the surroundings are much more beautiful than the pubs, which I usually go to. Thank you, Shantar, thank you, Jess, for inviting me and I must warn you that it has happened in the past once more, when I started putting together my thought in order to prepare the presentation, I realized that it had too many things to say, so I tried to squeeze as much as I could in the presentation, but I think that we need to have a second part of this conference in the coming future. So, the, I was writing last year, I spoke about big data, data science, and I feel obliged to explain a little bit of the background. As many of you know, I used to be an astrophysicist, and almost 20 years ago now, together with George Argoski, we slowly drifted toward, could you say George, slowly drifted toward the data science, because the amount of data which were collected by the new instruments was so large and so impressive that there was no way to end it with the traditional tools, and we began to explore the possibilities to use innovative software based on machine learning to easy the reduction analysis and understanding of these data sets. So, I became more and more involved in data science, and since two years I am chairing the data science initiative on my university, and I've changed profession, I'm no longer an astrophysicist, and I frankly don't know what I am now. I'm somewhere in the middle between different disciplines. Mainly, I'm interacting with engineers, companies, and building a new know-how on what's happening around us. And therefore, even with respect to my last presentation here in Second Life, much has changed, at least in my understanding of the topic and of the field. And this presentation, and if we succeed in organizing it also, the second one will be about taking away a little of the drama which has been mounting in the last two years around the topic of artificial intelligence. There is a hype around us about artificial intelligence. You know better than me that wherever you go now, you have artificial intelligence advertised. I mean you go to a supermarket and you find that the items were selected for you by an artificial intelligence algorithm. You go out and you buy a car with artificial intelligence. You go out with, I mean you take a plane and they tell you that the CIT assignment is done with artificial intelligence and so on. Artificial intelligence, in my opinion, is just a bad word. You know, one of those words which periodically are used by community, by journalists, in order to signal that there is something incredibly new happening. I'm not saying that there is nothing new happening. I mean, actually we are in the epicenter of a revolution which deeply affects the way we live and the way we work, the way we organize our lives. But from this to artificial intelligence, I mean, there is still a lot to come. So I mean, there is a lot. There are many considerations. At the moment I'm writing a book on the topic. So, you know, there is, my thoughts have not yet reached the sort of crystallized form. So, there is still a magma of ideas which come back and forward and therefore I'm quite sure that in the second part of my talk I will say something which will be in slight disagreement with what I'm saying now. But I mean, this is science. This is, you know, evolution of understanding and evolution of thought. Data science, as shown on the slide, is something very simple. It's a set of the principles, problems, definitions, algorithms and the processes which focus on extracting non-obvious and useful patterns from large datasets. Data science extracts patterns, not knowledge. Knowledge which is, say, can be assimilated to understanding is something different. I mean, at the moment we are still far from having an understanding performed by some artificial device. I mean, knowledge and understanding are still a human, by a human, you know, characteristics. I mean, so that's, and data science basically is what artificial intelligence is all about. So it's what I'm going to say for data science also for artificial intelligence. Our wisdom, Ariane, our battery wisdom, I mean, yeah. But the problem is that wisdom is a problem for natural intelligence, not a future problem for artificial intelligence. I mean, so basically the layman understanding of artificial intelligence comes out of science fiction books and science fiction movies. And this is also why I think this buzzword has become so popular because the artificial intelligence cares, commits, carries with it, you know, bad feelings. I mean, it is a scary term. I mean, people are scared by the idea that one day some sort of terminator or, you know, malvolunt, evil machine, you know, will try to exterminate a human being. I mean, this is a common understanding. If you ask other people in the streets, 95% of them will tell you that artificial intelligence will mean the end of the human race. Also, because the idea of artificial intelligence makes humans feel relevant in some sense, because if you think about it, I mean, so far, all the progress in the, in our technology has been, as concerned, only one aspect of the human being, the homophobes, the Latin Romans will say. The man as maker of tools, the man as maker of stuff. The industrial revolution was a revolution about capability to make things. The revolution, which we had experienced at the beginning of this century, was also saying technological revolution, industrial revolution, everything was about the capability of humans, of humans to make handicrafts, to make new tools, new instruments. For the first time, with artificial intelligence, we are facing an evolution which affects the homo sapiens, so the capability of human beings think and understand. And this is something very scary. This is something we have never dealt with before, and for which we have no antibodies. Basically, we don't know how to react, how to adjust ourselves. And therefore, there are huge social, religious, philosophical topics, which are under discussion, I mean, even legal topics. And there is a huge discussion going on everywhere on the, more or less in every community, this was a pleasant discovery. Before starting the large amount of research, I thought that artificial intelligence was a topic discussed mainly by computer scientists and mathematicians and by data scientists. This is not true at all. There is a huge community covering all aspects of human thought, which is discussing the topic from lawyers to judges to theologians, to experts of human science, psychologists, and so on. There is a huge literature far larger than any human being will ever be able to grasp in normal life. It's a very active, very vibrant, very exciting field in this moment, but basically we're discussing the sex of the angels, but people are so scared by what it could be that there is a lot of discussion about how to deal with artificial intelligence machines, which are not yet here. And there is a lot of confusion, and I think it's important to clarify this confusion in all the groups of people I've mentioned before, because they mess up between artificial intelligence and the wrong use of big data and of algorithmic capability to use this data in order to predict trends, to control trends and to manipulate the society around us. The real meaning, therefore, is that artificial intelligence is not in girls that a collection of different programs which try to address specific, simple, specific problems. So basically what the approach is, I isolate a specific process hidden within human thinking, like for instance, how do I visualize a vertical structure, how do I visualize a cat, and how do I recognize that it is a cat rather than a dog and so on. Then first you isolate the process, then you transform it into algorithms, and this is usually the most difficult part, and then you run it on computers. And then if you are successful, you start using these algorithms to substitute humans who perform repetitive tasks of the same type. For instance, already now in the United States, after the success of the automatic driving car, automatic driving cars and trucks, now there is already a community which is scared of losing relevance, like for instance, truck drivers. We have understood that within, let's say, five years, ten years, most of them will be out of the job just because automatic driven trucks will do the same job more effectively, less costly and way. So I mean, I'm not saying that this algorithmic approach to problem solution pattern finding is not impacting our society, it is impacting it a lot, and it will even more impact on our way of life in the coming future. The most unexpected way. Also because once you learn how to do the trick on one task, to export it to other tasks is much easier, you already know how to do it, you just need to do some bit of fine tuning. The only change is light and you always get confused with your controls. I apologize for that. Okay, should work. Yeah. One thing which I found very refreshing is that at some point I decided to, you know, to understand why suddenly two, three years ago artificial intelligence seemed to emerge outside of nowhere, just popped out suddenly. This word, if you look to a statistics of the usage of the word of artificial intelligence in non-science fiction literature, you'll find that there is really a discontinuity, before 2016, after 2015 and immediately after, because the number of times these words appeared increased by factor 100. And this is always very suspicious. And something suddenly becomes, you know, mentally popular. In little we say it doesn't smell right. So the first thing I want to understand is whether this thing was really something new or was just a new cover on something very much older. And I did as I always do when I am interested in new problem. I went back to the literature and tried to reconstruct a little history and I realized that actually there's nothing very new. It's what we call nowadays artificial intelligence, which for me is more or less a synonymous of data science is something very old. It appeared very, very long time ago from the fusion of two trains, which have been around since the very early stage of human evolution, data collection and that analysis. If you go to, because basically we are doing just with computers, something which has always been done. It's not in you. Data collection is as old as human civilization and the type of data are always the same. Already in the third millennium before Christ, there were transactional data, which means basically data about economic transaction, SLU, a camel, and they get so many cows and these type of things were already stored in some form, as well as non-transactional data. So data which were not really related to economic transaction. For instance, the Roman and the Gypsum sensors were no longer, and it was always a sort of transaction and they wanted to put, to tax people in a more effective way, so it always money was involved. But these sensors were built collecting words, collecting permissions about words, sex and this type of things. All this data were collected in common storage. So the collection of data is a very old thing. And the next step, the capability to retrieve data in an organized way, what we would call today structured data, is at least as old as the library of Alexandria in Christ. So I mean, basically the analogic version of a database. And this thing, therefore, since data collection is old, you know, it's very likely that also that analysis is a little younger. I mean, algorithms and simple tools to analyze and extract information from the data are at least as old. Here you have an example where you have the data analyst of the Egyptian time, which were the scribes, you know, checking the census information, applying a mathematical formula say, you know, in order to derive taxes. I mean, these things have been around for a long, long time and their relevance increased as the society became more and more complex. One big difference, this is a slide from my old presentation, is that in the past, to collect data was a quite expensive and quite lengthy procedure. Therefore, the boat rate, the number of data which you could collect per unit of time was very, very low. While nowadays, we are surrounded by digital sensor, by every aspect of our life is subject. Every time we use a card, every time we file an email, we send a Twitter, we post the submitting of the social networks, but, you know, every time we buy something, this information is automatically stored. So basically, data collection has become a very inexpensive thing. And the amount of data which is produced is just very, very large, incredibly large. The turning point, which were everything which we today, nowadays called data science, started this in 1991, which is quite a long time ago. It's when this guy here, Edgar Cobb, he was born in 1923, 2003, you know, put laid down the foundation of the so-called, what we call nowadays the relational database. This guy was a quite amazing scientist. He won the famous touring prize, which because it's a sort of a Nobel for computer science. He is also the man who invented the term online analytical processing, which is well known as all up now. And what is more important, he codified the analytical processing, which taught them of everything we are doing nowadays, which is 12 laws for, which must be followed in order to perform online data processing. It is nice to say that at the time he was a researcher for IBM, and when he published this paper, Relational Model of Data for Large Shared Data Banks, IBM didn't consider it at all. I mean, it just didn't take into account, and Cobb published a data to a third party, let's say, basically a paper which was sponsored by Arbor Software, later became Ipillian, and now it's a part of Oracle. And this is one of the main mistakes which IBM has done, despite all the many good things they've done, but IBM has done a lot of mistakes, because the guy who instead understood the potentiality of the codes, the paper was Alison, he's a founder of Oracle, and now he's the eighth richest man in the world. He basically, at the time the paper by Cobb came out, he was working for CIA, building the first database on codes principle for A8, but when he understood the potentialities, he just went out, started from CIA, and established the company, which was after the laboratory, which in 1982 became Oracle, which actually, in this moment is the monopolist of relational database all over the world. Also, they established the standard query language, which is data, relational database. I mean, this is important because it begins to tell you that what matters is who controls the data and how they control it, but we'll go back to this later on. The same part of the story, this was that election is data statistics, and the statistics, I mean, I found always very curious to remind people that statistics, the word statistics come out from state, from nation, from state, because at the beginning it was conceived as a way to collect and analyze data about the state in order to tax people. Once more, taxes are always at the heart of our life, for many years at the beginning, it was merely a summary or descriptive statistics, there was not much behind it because it was a way to characterize large distributed data using average properties, variants, some of the things which is very nice. The real revolution, which laid almost immediately to the foundation of what nowadays we call machine learning took place in the 17th-18th century, when the probability theory transformed the statistics into statistical pattern recognition, and there are many, many players in this story. So the first player I always like to put to this guy is Hieronymus Cardanus, Gerolimo Cardanus, one of the most incredible people ever existed. He was born in Pavia in Italy in 1501, he was a polymath, he was a physician, he was a philologist, an astronomer, a biologist, also an astrologer, at that time it was possible to be both an astronomer and an astrologer, but mainly he was a mathematician and inventor and look at this, he had also some problems, he was a gendu, he first believed in witches, he was a great supporter of witch hunts and farming witches on the stake, which is not very nice, and he was also a gambler, he was an inventor, he has invented the combination lock which you use nowadays on your trolleys, he invented the Jimmel, which is that suspension you see here in the middle, which is the three degree of freedom suspension which is used by gyroscope or mobilized surface, and even invented the Cardan shaft, which was the car, so quite a general guy, he was the first one to systematically use negative numbers, at that time they were quite new and he made public, he was not the discoverer, but he advertised very much the solution of other mathematicians for cubic and quartic equations, he also understood the need for imaginary numbers, and this is really quite an extraordinary achievement, but what is very important, in order to make good prediction for his gamble, he invented probability theorem, he's the guy who basically invented the binomial coefficient and the binomial theorem, which are a little bit, you know, the foundation of probability, and also he parted with the first theorem of probability, so it's nice to see, you know, that everything starts either from taxes or from gambling, not really positive, but I find it quite fun, it's the next guy, I mean, then there is a long, long succession of scientists established this new field, which we nowadays, since in the modern world, we like to put a label and we have a sort of a schizophrenic attitude towards things, I mean, we like to fragment our knowledge, you know, in domains, and the narrower they are, the better it is, because this satisfies our need for control, but actually, I mean, I find very difficult to establish boundaries between, you know, any aspect of human thinking, but so nowadays we call these things statistical learning, the statistical pattern, the cognition, the machine learning, or, you know, data mining, at the end, basically, everything goes back to these people. We had Blaise Pascal, who made some important theorems about probability theory, then the Swiss jackal Bernoulli, then Abraham de Moivre, who was one of the most unknown, he was one of the most unknown mathematicians, but he also is the one who established some very important mathematical theorem of probability theory, which enabled the estimate probability distributions. And then Thomas Bayes, I think that most of you have heard at least once Bayesian approach, Bayesian statistic, Bayesian bias theorem, and so on, is a very accurate way to infer knowledge out of data, a good way to extract information from data in presence of partial or not information. Actually, Thomas, I never understood why the bias theorem is called bias theorem because he had the very, if you read this paper, it's, he has a very big bad understanding of what he wants to say. Then this guy here, Richard Prince was the one who began to put his, Thomas Bayes ideas in a more formal way, but the real inventor of the statistic was an astronomer, Thomas Bayes, over there, Simon de Laplace, one of the most influential astronomers of his time. He took the intuition, the very rough intuition by Bayes and developed it in, but nowadays he is known as the bias theorem and he established the Bayesian statistic. With the foundation of Bayesian statistics, we are basically beginning, so it's already the beginning of the 19th century, it was in 1805, we have all the bricks which are needed to do statistical pattern recognition, which is one of the main aspects of machine learning and data science. So these tools, these algorithms, these mathematics have been around for a long, long time. Then what has changed is the way we have implemented it into computer, but we go back to this in a quite short time. Now, another good guy is Carl Friedrich Gauss, who contributed a very total piece of information, you know, because basically he is the one who established the theory behind the regression, the logistic regression, which for everyone works in machine learning are very too familiar terms, but the mathematics behind them was established by Gauss in the attempt to find the elements of the orbit of Ceres. Ceres had been discovered by Piazzi in the 1st of January 1801 and Carl Friedrich Gauss solved a very difficult problem of finding the orbital elements of a new planet using only three observations. And in doing so, he invented the least square method, the method of the minimi quadrati, which is most common tool nowadays for replational data. And in doing so, he established the theorems behind linear regression logistic regression. From the beginning of the 19th century, all the bricks were in place. Then there was a long time of quiescence. In the meanwhile, there were significant advancement. For instance, nowadays, there is a lot of talking about data visualization, advance of the data visualization, which is correct. But again, the problem is a new one. It will start, everything started actually with this guy here, almost unknown to most young players. He was an English engineer and was working for Ray companies. And he developed the statistical visualization. He is a guy who invented the area charts, the line charts, the time of Instagram shown in the bar charts, the pie charts, all the tools which he used nowadays and which is somehow related to the field of mathematical visualization, which is moving to the advanced visualization we use nowadays in computers. Obviously, things have changed a lot thanks to the introduction of computers or monitors. Here you can see modern visualization of complex data, which would have been impossible over time. But remember that in most cases, more complex does not mean deeper. It's just that you are using old tricks to solve new problems. And you have a much larger computer power, which allows you to do much more complex tasks. But not much new under the sun, I will say. Then there are exceptions. For instance, you can visualize things in many different ways. And I think Jess remembers our first experiment in second life, George Raffaele, Gino, myself, were using the second life to visualize complex data. I had a sample, but you know, no, there is not. And I must tell you that this is something which should go into the curriculum of second life is that Ciro Donald and George experimented in second life and kept working on it. But basically using the same philosophy which was behind the debt with Jess's life. And Ciro Donald has started his own companies, which is nowadays one of the leaders of the visualization in the world, which is in politics. If you Google it, you're impressed by what you can do. Basically, it combines what he tested in second life with the virtual reality, augmented reality, uses HoloLens, Oculus Rift, and so on. You know, to allow you to interact with the complex environment. And this is, I would, I like to say that this is a spin-off of research done in second life, 10 years ago or more. Okay. Third piece of the bridge of the world is everything. Basically, the mathematics was in place. There has been not too much innovation in it. Also, the emphasis which has been recently put on deep learning, deep learning has been around for the last 20 years. Overcalled ensemble learning and so on. The only thing is that now these networks, which were all fully slow when you were running them on single processors of the previous generation, now can be running the cloud on a huge amount of data. And obviously they lead in many results. But in deep learning, there is nothing new. This is something which is seldom spoken out loudly. Google has invented nothing. All the theory, all the, all the mathematics behind deep learning was already in the machine, in the data mining literature 20 years ago. They just, you know, optimised the algorithm they invented, specific sub parts like adversarial network, which are, you know, but so specific problem. But behind deep learning, there is nothing new with the exception of a huge amount of computational power. So the revolution started, this is just for historical reason, with Alan Turing who invented the first computer and things, you know, exploded after that. This is again a slide from previous presentation. Nowadays, it is estimated that there is at least one processor every, we have a huge amount of processors and we have a huge amount of computing company, especially if you connect this processor with the network cloud computing and things like that. Ah, two other people, which I really think are always overlooked. I mean, who are really instrumental to what is happening today are the old Shannon, who really is the father of modern information theory and also is it the AORM or the Shannon theorem is it the central or it was in American cities and to publish a terminal paper and mathematical theory, one of the forces or whatever happened after. And then this lady, which is, who is really very unknown to most people, Erwin Ficks and their student, Joseph Forbes. Well, Erwin Ficks in already 1951, we are talking about 68 years ago, invented the AMN, the nearest neighbor method, which nowadays is one of most used things in data science. And one thing which I really am sorry for is that nowadays the way science works, which I mean, I'm an old guy, I'm 62 and I've been in business for 41 years now. And when I was young, people read literature nowadays with the attitude of modern science, which is publish or perish, basically you have to publish. It doesn't matter whether you publish good thing or bad thing, you have to publish because it is based on the number of kilos of papers you're publishing. Even if that paper is worthless, it doesn't matter. But basically people publish, they are in hurry to publish, they basically, they don't spend the time in reading, they don't spend time in thinking very often. And what happens is that, for instance, PhD students or young researchers do not read any literature which is older than five years at most, not usually three years. So you find nowadays that nearest neighbors is attributed to some young, not so young, but to some American researchers who republish 10 years ago. And the name of all of the fixes is virtually unknown, but I invite you to read the papers for those who like these things. It is the nearest neighbor, she invited it completely, and she has never been a knowledge for it. And she's rightly to be considered one of the founders of modern science, or modern data science. The time of traditional literature as from a scientific point of view, intelligent robots have been in the science fiction since 1852. But comes out of a meeting, like, you know, another thing which is surprising, this innovative word, these bad words always come out from meetings. For instance, the famous Drake equation, which is used very often to estimate the number of extraterrestrial civilians, the probability that some extraterrestrial civilization is two years ago. Drake equation and artificial intelligence come out of workshop. Of the planning for workshop or out of, in the case of the Drake equation, it comes out of the name of the session of the first meeting on the first for extraterrestrial. There is nothing simple. It's just, you know, a sort of summary of the session or workshop. This guy, John McCarty, one of the amazing minds of this century is a post-seconding medalist and mentor of the least. In 1955, proposed to the short side foundation, a workshop to analyze the possibility to teach computers or to program computers to performance tasks, which at that time were considered to be an exclusive prerogative of the human. And in that paper, for the first time, appeared the word artificial intelligence. And with co-authors, co-authors of the proposal were Kuroschans, and Marminiski, and Pater Moshesler. If you read his motivation is quite impressive. This guy then moved ahead. I mean, he invented the first language for this type of application, which is LISP. Then he contributed to Al Gore. I mean, he died a few years ago, but he has been active for the last minute. It's life. Quite impressive guy. Okay, so let's go back to the beginning. I mean, it's a long history. There is, as it always happened, most of human endeavors, there is no discontinuity. There is a slow evolution a better understanding of what's going on. And then suddenly in 2015, as I said before, artificial intelligence emerged from the science fiction. A very common phrase, a very common word. So why? Well, I think that now it brings back to the fact that for the first time, all things which have no intelligence whatsoever together, I mean, big data, no intelligence, no instinct intelligence, computer networks, well, less than less. Computing intelligence is just a set of algorithms, which are doing what I've been telling you so far. And they are also not intelligent. My dog is a magnet. And then the intelligent human machine interfaces. All these things together conspired to make the world artificial intelligence credible. Now it's important for me, and I am not a solution. I mean, this is an old slide. I am not a solution for this one. To understand who really is behind the wheel. I am my suspicion, but I'm not saying that we are not going to work in a far future toward an artificial machine, artificial intelligent machine. It is possible. It's feasible. It's just matter of complexity. It will take long, probably, because at the moment we are beginning to understand, it will take the convergence of many science, biology, biotechnologies, machine learning and computing and computer science there. But it's still far to come. So why suddenly artificial intelligence has become so popular? The data. It is a huge amount of data. And I know that now I'm going to repeat something you just said last year, but I want to say it in a different way this time. Because I have a different perception of the whole scenario. Artificial intelligence is, in some sense, emulated in specific tasks, which still have some algorithmic complexity behind it. But they are still in some way deterministic. They are still algorithmically defined by algorithms, by world-defined sequences of commands. For instance, computers now usually beat humans in chess, in chess playing, or even in television quits like Geopardy. They drive cars, like they land planes, they land probes. But if they have very complex problems, with some amount of determination, but everything is coded through algorithms, through a very smart exploitation of data collected by humans. And the algorithms are written by humans to emulate data-specific behavior or their brains. So it's something very different from what we all envision as artificial intelligence. The problem is that this revolution, for the first time in history, is not controlled by government. For anarchists, this can be good news. But for humans who believe in democracy, this is very good news. And they think that also the sudden increase in the use of genoterm artificial intelligence comes out, and also the emphasis on the fact that artificial intelligence is coming, artificial intelligence is behind the corner, comes by reasons which are very little to do with science. Another investment in the infosphere, where the infosphere is a name invented by the director of the digital ethics lab, Oxford, Luciano Floridi, is a complex of net plus infrastructure plus software. Basically, it is one thousand giga, one trillion dollars in less than ten years. A column of one hundred dollar bills which goes from the earth to the moon, given the idea of the money which has been invested by government and states. Then you find that the largest company existing in this moment in the world, the three companies which have gone beyond the one trillion dollar of corporate value are people, Google, and recently the Chinese giant Baidu. They are all related to the data. If you see, this is the Google galaxy. At some point Google is, I need to change my position because I cannot see something, just a second and it will work. I cannot read the number because my label falls, okay. It is 1.65 billion dollars for YouTube and it pays 19 billion dollars every year to control the traffic on the apps on Android. Now, one thing which is a very often let me go ahead for a second, then I go back to some consideration. They look for instance at Facebook. Two thousand two hundred billion users, so basically two, no sorry, not billion, million users, so basically 2.2 billion users, which the same Zoukemberg calls the use of a video. This means something and as a value of 700 billion US dollars and there's both for Instagram for 741 million dollars with only 20 million users, while I say it will stop 19 billion dollars with 1.5 billion users. Microsoft, 800 billion USD values, 1.5 billion users buys Skype by paying 8.5 billion dollars for 633 million users. If you look at it, you see that there is a direct correlation between the number of users and the value of society and also the fact that these people are trying to create conglomerates which monopolize the data. Okay, why are they useful idiots and why are, why it is so? Where does this value come from? I like this picture because here to see useful idiots usually is used in an ironic way, but here you see real idiots using, providing big data to the big data provider, you know, while driving, while crossing the street or while in the classroom. These are the traditional idiots to which we all are used, to those who have existed since the Babylon time, you know. But for Facebook, for Zoukemberg, all users are useful idiots because basically they provide the data for free without realizing the huge value which is associated to this data, which they are providing for free. And this is also why these people rule the world. I am getting close to the end because I see that they have run for a long time. It's in five minutes and done. So basically, what is happening in this day is that at the beginning, I mean, there was a network. The network had many, the web, the only web had information in it. This information had to be found. In order to find it, a few companies built the first browsers. These browsers were exploring the network continuously and indexing the contents of the network. So, you know, the first Safari before that, Netscape and many other browsers, then Google came among these many players. In order to store the information, you need to build infrastructure. In parallel, in order to index this information, you need efficient algorithms for machine learning, you know, which basically go through the information that has been found on the network, you know, a code and convert it in keywords, which can be indexed and then searched by the browser. And in order to do this indexation, you need computing power. But again, what happened is that by chance, the people who started the first browser almost as a game found themselves in a virtual loop, basically in a spiral. More nodes in the network, because there were no users, there was more information, more information to be indexed, needed more storage for the indexation and more efficient algorithms for the understanding and classification. In order to dissolve, these companies needed to acquire larger infrastructure and so before the governments could be realized, could realize what was happening, these people become at the monopoly of the game. They basically became the owner of the infrastructure. Now, where does their huge richness come? The big data provider, you know, big BP. Nowadays, they own the world, please. They are by far more powerful than any company, than any large international conglomerate or companies. They really own the world. Data comes from the data man, not because as many people mistakenly believe, you know, the data used to sell advertisement, not at all. The advertisement for this company is basically peanuts. I mean, you won't advertise a new type of shoes, okay? You pay one million dollars, it's noise for these companies. Where does the richness come from? The value of this company, by the fact that they can use the information they have in their data set to exploit the society around themselves. So basically, they can sell this information to companies in order to improve their market. They can sell services for specific topics and so on. And this situation cannot be resolved. Now, since these companies are behind this machine learning and artificial intelligence, you realize that the main drive behind the sudden success of this artificial intelligence warning is related to the role which the data provides and conquer it and very careful not to lose in the normal world. I mean, well, I'm sorry that it took longer than foreseen. There are many other things which I would like to see, but as a lapse, that's all I would like to finish here. If there is any question, any plan for the second part of this talk, I'll be very happy. So, let's finish here. I don't see any reaction from the public. Here we are, the shot. Okay. Thank you very much. Thank you very much. I'm always glad that we're such an audience. You have an equation. You know, I always found that the three levels of robotics are not really believable because I think that let me put things like that. Asimov was writing this thing in a pre, in a racist and era, basically. When there was still the understanding that some types of humans or some races were better than the others, I'm not saying that he was a racist, but basically he was not. But the culture of that time was permeated by this behavior. So basically he created this sort of artificial intelligence, which has no freedom of choice. Basically he created the slaves. The rabbits by Asimov are are intelligence with the limited range of activity. And this was basically these three laws are dictated by the fear that humans have toward people who are different. The same reason why we have racism, you know, again, the people coming from other countries in the United States against Mexicans or, you know, they are different. Everything which is different creates fear. And we need to build the walls. In the case of Trump, in Italy, we need to keep them away from the coasts due to Salvini or, you know, in the case of Asimov, you need the three laws who are needed to protect humans from Robo. I think that an artificial intelligence will be exactly what the name says. It will be an artificial intelligence. She will not hear at all about any laws we can put in her brain just because it will be intelligent, it will be free, it will be able to decide by itself. But this is just my personal belief, you know, belief. This is a long question and I have to read it. Just a second. You know my sincere answer. I think that unless we do something, we do it at the governmental level and unless we recover a sort of popular control, state control on big data and on the usage of this big data, we are bound to lose our freedoms, our civil rights, and basically we are going toward one of the worst forms of dictatorship, which human being has always experienced. Also, because remember states right or wrong, dictators, non-dictators somehow want the good of their subjects. Private companies know. McZuggenberg doesn't care about your well-being, doesn't care about your freedom of society. Only thing he is interested in is profit for his company. So for sure if we leave a big data and the only revolution in the hands of big data providers, we are bound to form all the tyrants. I cannot read, sorry, yeah. I will go prospect, I'll give you a video, yeah, definitely. But I mean it's, you know, usually I'm a very anti-government person, but in this case I would say that government have no, no responsibility for it. I mean governments are things which react very slowly and the evolution of this field was very quick. Basically they were found unprepared, they had no time to react. And now the situation is rather big because of the control of the data, the control of their distribution of their usages in the end of a few tycoons or not subject to any formal control. And this is scary. I lost the food, so I can comment on that. I mean there is a cool nursing later. Sorry, I'm in problems with this computer. Okay. But there are many, many studies about the way robots should look like, you know, when they, if they must look like human beings and not look like human beings. Now to tell you the truth, I have no personal idea about that. I mean the thing is still far away in the future, but people are just to everything. And so I really have no answer. I think there is a problem behind it, but really no answer about that. And so basically no answer. I don't know. A comment to Taglin. Taglin, I don't know Taglin with you because the motivations behind the government, you can, you know, I am a democrat. I don't like, you know, many republican ideas. I mean, the equivalent in Italy. So in Italy I am on the left side of the world, rather than on the right side. I don't like the right side of the world. But history teaches us that government, even one government can always be corrected. We are the fascists and we took it down. There has been communism and we took it down when it didn't work. The problem is that when we put things, I mean, why there's a scenario, but now we go away. Well, I know you're from the middle of science, the opinion. I think that one of the worst problems nowadays is that the advent of a big data artificial intelligence is taking place in a vacuum or ideologies. In the past, basically for the last two centuries, human beings have been driven by three dominant ideologies, fascism, communism, and liberalism. Fascism was basically taken care of by the Second World War and we were left with communism and liberalism. Communism has been taken care of like it or not in, by the world when the Berlin world came down and liberalism has been taken care of by the crisis in 2008. So for the first time, we are facing an apical change in society, an apical revolution in society without any ideology. And this is terrible. Because if you don't have ideology, you don't have a path to follow, you don't have goals. You don't know what you want as a society, where you want to go and now to achieve it. And therefore, we are living in worse conditions. So maybe it's very different to have a bad usage of technology or things from a government or from a corporation. Governments can be somehow controlled and corrected. Corporation not. Thank you. I also must go because for me, it's dinner time. For me, it's, it's very late actually. It's 8 20 p.m. In any case, I mean, if you wish, I can give the second part of this talk in, let's say, near future. I leave to Chantal and Jess to decide. Thank you very much to all of you. It was really a pleasure to be here. Have a good night or a good day, according to your time zone. And most importantly, a good weekend. Shout out to everybody.