[inaudible] that we are having this conversation today and we will try to get back to you is the tenyear anniversary of the debut of the iphone. With that in mind, we are pleased to be welcoming back andrew mcafee. We hosted them for their second book and now they are back tonight with a followup of sorts and that book is the machine platform crowd the Digital Future which we have for sale following the conversation. With a further introduction, the Research Scientist and director on the Digital Economy at mit and both of them on how it is changing business economics. Joining them on stage tonight is a reporter where she covers the start of economies from the magazine where she wrote about business economics for the funny box column. We have copies of the book for sale following the conversation and they will stick around and autograph them for you and without further ado. [applause] thank you all for coming, start by telling us a little bit about your book. The book is available for sale [inaudible] [laughter] its about the great rebalancing we couldnt fit all the words on the cover so we used the last three and that is the direction the world is changing right now. Its born from the human mind is towards machines, datadriven and Artificial Intelligence moving towards platforms like the five most valuable companies on the planet. They traditionally made the decisions towards the crowd. As we heard during the introduction we wrote a book that came out in early 2014 called the second machine age and it was about this progress that we are living through and as soon as we published that we noticed a phenomenon that we talk about the book and then we would have interesting hallway conversations over and over again somebody would come up to us and say hi believe the story you are telling. Now how do i think about my Business Model and the industry that i am competing. I know things are changing and i need kind of a guidebook for this. But the more that we realize this is going to be the next book we were gone, the central problem was we didnt know the answer to that question when we first started hearing it over and over but we turned it into a feature and said this is intellectually rich territory to go out and write a guidebook for this new economy for this Business World we are heading into sovie sawyer doing a lot of research into digesting papers and all that. But what we do that i is littler different and much more fun, as we go out and talk to the alpha g. s. They are changing things quickly so we want to hit the road and talk to them a lot. To come up with a framework and think about the changes that are happening so quickly. We couldnt sleep out the revolution we couldnt see this industry after industry and they all look like platforms to us and we kept seeing these companies and individuals hand they were activating very Diverse People around the world it was a powerful phenomenon but that wasnt a threepart structure in the buck. One thing that comes up in the topic you had earlier in your book is a sort of good way. Here is familiar with the alpha story . Do you want to explain what happens . It was trained against the ancient game of go that is a little boardgame that is vastly more complicated than chess which was made by humans back in 1997. Though, there are more possible moves in the game than there are in the universe and that means a traditional programming techniques of trying to search exhaustively, even just tiny fractions of the move is completely inconceivable. As some of sunday against no they didnt just last year and they did it by the techniques not to exhaustively search the move but by looking at the board and seeing this pattern. If they cant explain how are they going to profile is with this new approach to the Machine Learning with looking at lots and lots of examples. When they see what the machinese doing they say we havent even touched the edge of the game. Weve been studying this for 3,000 years but they are so far ahead of us we realized we have no conception of what is possible. Despite the fact we read about it in the press it is for three millennia for 3,000 years we built this body that was exhaustive knowledge about how to play this game. This piece of Technology Comes out everywhere and shows us how much more headroom ther head ron this particular domain. When we think about all of the other incredibly complex things we are trying to get good at there that is solving the mysteries of cancer or unlocking the secrets or what to interest in or whether it is dealing with the complexity of the human genome, we now have a colleague for that word and that cannot be limited by the knowledge that we possess right now that might be able to show us some ways forward. Sometimes we call it the second wave of the second machine age. It is the idea that machines start doing these tasks in the way that the first machine started doing the physical tasks. But the way for the past few decades is we programmed them and codified knowledge should havthat youshould have to tax pn were a spreadsheet and we started doing exactly what needs to be done. Into the machine what to do the same thing. Thats great. The generated petroleum of dollars worth of values o valuet is great that there are so many problems it sounds paradoxical that we dont really know how to do them. I would know how to code, how to recognize my mothers fac motherr ride a bicycle. These are just mean no more than we can tell. In the Machine Learning we have learned how to open up those possibilities. They are learning on their own how to solve the problems not teaching them painstakingly but instead by giving them an example of what works and what doesnt work, this is a example to figure out the patterns on their own and its a remarkable transition there was an example that went viral. Can the machines tell whether it is a blueberry muffin or a dog. Of their most elaborate of google or the kentucky fried chicken. And not getting it right all the time it would be hard to tell apart but we can often do it and what is really remarkable is thatheheadthat they had about ar rate and nobodys dow now it isr than 5 which is less than that on the same so we kind of croft displayed its not just a matter of the degree that when you raise the temperature of water passed a boiling point it is a fundamental transformation. It is a machine to do a task better than a human than if you are an entrepreneur and manager and have to make a decision you are going to pick the one that can do the job better people would be learning and after the World Champion was defeated, he said so beautiful and that is something we could see him a lot of these fields. You are right about people running sidebyside with these machines. Itits part of the reason that there was a downward blip. We think two important things happened along the line of the question. One is that they were beating about 70 of the opponents and after he got done playing and dealt with that technology for a while he was making a big quantum leap upward. The other thing that happened skyrocketed and they started playing at a higher volume than they were so isnt it terrible that we are no longer the rest of this activity and doesnt reduce the human value or dignity. I think that is wrong because if we define ourselves by the things that is presumabl silly do. Let me underscore one of the planes that is made. This came in a fundamentally different way. For 3,000 years theyve been learning from their masters if they made that it would slap the wrist and say dont do that. But then it came at us with a fresh perspective and they had a different way if you want to call it that. Its most importantly that this is different in certain ways and come at this with a different perspective and that opens up the possibility to Work Together and help each other because they are not just at the same thing they are two different kinds of intelligence. You need to b it to be good t this domain to show why it was behaving in this way because the machine was kind of amused about why it was doing what they were doing. We need higher level humans to look at that and think i know whats going on here lets test that idea. Where are some companies parts of the economy that we are already seeing instead of this replacement with a mutual collaboration of learning . If it is in the spirit of the book one of our friends runs your custody but has a bunch of courses you can learn Different Things in 90 days so people come to that website an the website t with sales reps. So some of them are better than others and are able to answer the question effectively and convince people is this a good course for you to take. Others, not so much. And along with their students if you take veterans transcripts that becomes the data for the supervised learning that is giving examples of success and failures and letting the machine figured out what matters. So they get these transcripts and we start seeing patterns that they have different ways of answering questions. And at that point, they do not turn over because there are too many unstructured questions that there is a sort of 80 through 20 and they still have the humans doing interactions with the incoming traffic but when the question came up they said here is a phrase you might consider using in the past. They would pick up on that and they sold their improvement dramatically and it went up by about 60 said it was a tremendous improvement par but t of it is the ability to combine humans and machines in a way that neither of them would have been successful on their own. In some cases we are going to automate work and thats the appropriate thing to do but its naive to think this is the only direction or the only force going on. We loved that story because it shows we are fans of technology and automation and we can kind of amplifying and turbocharged their ability with a lexmark technology in the background. We are not saying turn the dial to one extreme. This question comes up a lot what do you see in the areas that are more distant to the advent of machines and Computers Like we were adopting earlier where you mentioned this famous saying wha would you like to que it . He said a while back the only difference between the series of observations is the excessive confidence on the part of the song writer because it would be trivially easy to program a computer to spit out observations that might have words associated. The lyrics probably wouldnt make a lot of sense and wouldnt activate any emotion on the part of the rest of us and so why would they do this when they have this observation in the way that resonates with other human beings and that is a skill we have not seen Computers Come up with yet. To see whether that is kosher or not, i speak some french come i cant do that in french to save my life come english as a second language speaker then its any other native speaker. The reason i bring that up is i believe that we have the native speakers intuition about the condition of the social world we have created in the physical move between world through trying to teach computers and if we do that it is decades away in the same way its a second language speaker that weve created so i have a lot of confidence that the human ability to move forward in this response was and the language im actually going to be surprised. Its hard to tell them apart from the musical components composers. So what is the difference between the music and lyrics. This is getting away from my area of expertise but we know what chord progressions sound good and like sirens clinging onto the new york streets and we can encode at that stuff an theu know the rules of a feu fugue oe things it is relatively easy to hit the button. They are not necessarily encoded with a few. But there are patterns that emerge and you dont know how, dont have to know anything about these patterns and they are quite good at these that for the year x. You need to understand the meaning. Its not just fo that the two ws often appear with each other. You need to go beyond something called the human possession. And although they could recognize the tears they still do not really understand what its doing. They can translate one language to another but they are not actually understanding what is going on. They can be very good at certain kinds of video games but if it involves a deeper planning instructor, they are hopeless at that and i think the language is a little bit more and music is like the power match. We included in the book an excerpt from a machine that got trained on the jane austen novel and it actively makes your head hurt. It is not pleasant to read. Its often said when you look at jobs and what is likely to become automated then its not that high skill and low skill as much as it is repeated and highfrequency versus variable unpredictable tasks. The broad lesson of the moment is machines are not just good, they are superhuman at certain types of tasks. A there will be involved more complex planning those are things that are much harder to have at least with the additional curve. We are talking about what is available on all sorts of things that made the breakthrough 2018 we are not sure but based on the correct Technology Committees seemed to be taking a bunch of inputs. I think it is becoming out of date because i never would have thought of the medical diagnosis in the work that you would hand off to a machine. Its an incredibly important and difficult and subtle complex exercise. Machines are better at those things. I think the medical diagnosis is squarely in the sights of these automated tasks and one that we should apply a lot more machine to that it doesnt appear the case five or ten years ago. You can also apply it to the medical images. They spend years of training to learn what to look for you train the machine to do that and they can very rapidly learn not only all the markers that are taught in medical school but a number of new markers that are never noticed before. What is the accuracy you talk in public about how they are rapidly equipped. This is a number that is changing as we speak and there is a significant article in february of this year going from 30 to 5 that was an example in the Voice Recognition it went from 8. 5 to 4. 5 but what is striking isnt just the improvement but it was done over the past ten years and past ten months since july the a lot of it comes from the broad set of algorithms but a lot more data and computational power. And at th that the studies we come across or conservative but they generally say that a properly done Machine Learning system will be at the human level performance. I think they are better than the average doctor in a lot of the cases today. Where are people going to really encounter with this kind of technology, is at alexa by eating whole foods . One of it is Voice Recognition and its not perfect yet it still makes those kind of errors that you described. I now dictate a lot of the text and there was a study out of stanford thats about three times faster and more accurate. Weve been encountering some of them are the leading example that has been possible has been deciding the searches and if they can figure out what you are likely to click o on and what youre not likely to click on that is worth billions upon billions of dollars and one of the things that is driven some of the algorithmic improvements. We are throwing all of this brain and computer power and at the insight we get from that so the main way that we are coming against these technologies is that the viselike said speaking into them and the letters and figures of what youre looking for those that use facebook as of yesterday it will offer the labels and recognizes who we are in these technologies something we could do just a couple years ago. Just me personally i find that creepy. The first time that somebody else tagged me in a photo and put that on facebook was years ago. It is a creeping boundary and i think thats turned fairly or very quickly. The analogy that i use this when i was learning to scuba dive, the first breath i took they told me you arent supposed to breathe underwater. The second one is any advanced technology is indistinguishable from magic and maybe that is a little bit creepy but the companies that are pushing them out if you think a little bit about how people are going to react and there are certainly examples where they cross that line and reveal that they know more about them than you wish you knew. By exposing that they still know it and it is one of the things we have to think a little bit about is what can these algorithms infer about us from our behavior and who our friends are and about social networks. All these things are providing data to the Machine Learning algorithms and whether we know it or not, whether we like it or not, they can potentially be used to infer things about us. On the other end of awareness there is the self driving car issue where they are moving too quickly, its going to cause an accident. It will cause more death and regulators to clamp down the whole program. The half a view on this which is the self driving cars are not going to be perfect anytime soon. The right standard isnt or shouldnt be perfection. As you may know there are about 30,000 per year so even if they are 90 better, that would be 3,000 deaths and we saved 27,000 lives but the reality is the self driving Car Manufacturers are probably not going to get 27,000 thank you letters devoted 3,000 subpoenas. So this is something that goes back to the cultural changes we have to think about the expectations and where we would like that threshold. Another one of these scuba diving moments or terrifying. This was initially not cool at all. It was going down the highway and it was an amazingly boring experience. Mine was basically fine. But the New York Times reporter, hi, his car did starto he got the best lead for his story. I am in my car and pressed the start button and it wont turn on. So this whole question about autonomy is important. Technology right now is increasing the utilities, the numbers are going up. This is the worst of all possible combinations. I view is for heaven sake lets bring forward the day that we can turn this over to the machine and they can do a better job and think about that a little bit to the elderly, brian, blind, disabled. What happens if you have a self driving car coming down the road and someone just decides to walk into the road and stand there how does the car negotiate that . The car will stop. Does the pedestria