>> Amy Draves: Thank you for coming. My name is Amy Draves and I'm happy to welcome Erik Brynjolfsson and Andy McAfee to the Microsoft Research Visiting Speaker Series. Erik and Andrew are here to discuss their book, The Second Machine Age, in which they reveal the technological forces behind an economic sea change. Erik and Andy are cofounders of MIT’s Initiative on a Digital Economy, and Erik is one of the most cited scholars in information systems and economics. And he is a principal research scientist and the author of Enterprise 2.0. Please join me in giving them a very warm welcome. [applause] >> Erik Brynjolfsson: Thanks very much. I'm Erik. That's Andy. And we are delighted to be here in part because as we were writing this book and as we continued to do our work we really benefit from interacting with audiences like this. We actually did some of our research here at Microsoft Research to understand some of the trends that were going on as well as courses and labs at MIT, computer science and the i Lab, the media Lab, folks in Silicon Valley and we spent time talking to a lot of economists. To be frank, we ended up kind of confused and that was part of the origin of this book. Confused, because there are so many disparate trends going on. In particular, when we talked to technologists like many of you are, I'm sure, we often come away amazed at the things that you're accomplishing and that not only are already being done, but are in the pipeline that we're likely to see in the coming years. There's sort of this common optimism, infectious optimism, or even techno-utopianism that you get when you talk to technologists about being able to solve all sorts of problems and that the world is going to be a much better place. It all seemed to make a lot of sense to us, but then there are other groups we talked to, economists, for instance. I was just at a meeting with the American Economic Association in Philadelphia where I was on a panel with three other economists and I realized there's a reason that economics is often called the dismal science. They have a very different perspective on where we are and where we're going. They have a lot of data behind that as well. Economic statistics in some dimensions are actually pretty dismal. Median income today is lower than it was in the 1990s and, as I'm sure you know, employment has been falling as a share of the overall population. Unemployment is still quite bad. To the extent that it's improved a bit, that mostly reflects that people are dropping out of the labor force, so they aren't looking for work anymore, as opposed to a lot of new people being employee. Andy and I sometimes call this the great paradox of our era, the fact that on one hand innovation has never been faster, yet people are more concerned about their futures and their children's futures. We have record levels of wealth in the country. We just hit a new all-time high, $77 trillion a couple of weeks ago, record levels of GDP, record levels of productivity, record levels of profits and all those numbers are going up, but not everyone is sharing in that. If median income is lower that means that 50 percent of the people have not gotten better off, and many of them don't even have jobs. So reconciling these two disparate trends is something that we really struggled with and we were confused about and as we were working on the research for this book we ultimately came to the conclusion that paradoxically both of these sets of trans actually have a common cause and that is the way technology is being deployed and unfolding in the economy, not just in the United States, but worldwide. It reflects an important fact and kind of a dirty secret in economics really, which is that technology tends to grow the economic pie making us wealthier than ever before, but there's no economic law that says everyone benefits evenly from that advance. It's entirely possible for some people, potentially even a majority of people, to be made worse off, not just in relative terms, but in absolute terms. With earlier ways of technology, it was a relatively small group. Buggy whip manufacturers were made worse off by the introduction of the automobile, but in time everyone benefited, and that's been the case with most other major technological innovations. But if you look at the data that hasn't been true more recently. The latest wave of technologies seems to not be benefiting people nearly as broadly. The data show that for the past 200 years productivity, GDP, median income and employment were all rising more or less in sync with each other decade after decade. But starting really about the early 1990s and accelerating since then, the trends have started diverging. Productivity has continued to grow pretty strongly, as has overall GDP, but median income became disconnected from it as did employment. The nature of the new technologies is part of the story behind that. We call our book The Second Machine Age, which is an obvious echo to the first machine age, by which we mean the Industrial Revolution. Throughout most of history, economic prospects for the vast majority of people hardly changed at all until the first machine age when we hit an inflection point and growth really took off. That was in the late 1700s, early 1800s. In the first machine age, for the first time we really learned how to automate and augment, remove the limits really from our muscle power and provided much more access to physical power through a whole set of machines, the first one big one being, of course, the steam engine, but later the internal combustion engine and others. That sent living standards soaring, which we're still benefiting from even today. What's the second machine age? The second machine age is when a set of technologies is affecting not so much our muscles, but our brains, removing the limitations from a lot of cognitive functions. Like the first machine age, the second machine age is unleashing enormous wealth and productivity and raising living standards on average, but it's also different in some ways. You see, in order to take advantage of physical power systems, you need a control system. You need brains to manage the physical power systems. As technologies like the steam engine gave us more physical power to manipulate the world, it was actually very complementary to human labor. It required more human workers to manage that and that was a synergistic system. In the second machine age as we remove the limits from a lot of cognitive tasks, it's not so clear whether humans will be complements or substitutes for those technologies and that could have very different effects on labor and on the distribution of income. Let me give you a little bit less abstract and stratospheric about that and talk about some specific applications. One, for instance, is an event that happened just as we were beginning to think about this book. I was getting onto a plane and the person in front of me was kind of talking really loudly on his cell phone and I couldn't help but overhear him. I'm sure you've been in that situation before. And what he was saying was oh, no, no. I don't use H&R Block anymore. I use TurboTax; it's faster, easier, cheaper and so I don't need a human tax preparer anymore. And I was thinking about that and, you know, of course he was absolutely right. TurboTax, $39, can do your taxes very quickly, efficiently, accurately and it's hard for a human tax preparer to compete with a $39 piece of software. What Intuit has done is, I think Microsoft has a product in this category too, doesn't it? I'm not sure. No? But anyway, Intuit has taken a process and codified it and then taken it and digitized that. And once they be digitized it, as you guys know better than just about anyone else in the world, you can make copies of it and you can make ten copies of it. You can make 100 million, or a billion copies of it and each of those copies is a perfect replica of the original one. Each of them can be made for almost zero marginal cost, and each of them can be transmitted virtually instantaneously anywhere in the world through the internet. So they are free, perfect and instant. Those are three characteristics we hadn’t seen in goods and services in most of human history, but those are the characteristics of digital goods. Those technical properties lead to some interesting economics. In particular, it's very common in digital markets for them to be winner take all markets. After all, if you're buying a piece of tax preparation software or any kind of software, it's unlikely that you're going to want to buy the tenth best program or the thousandth best program. You're probably going to want to buy the best one that's available and so there tends to be a great concentration of the revenues in just one or two or maybe a handful at most of programs, instead of tens of thousands or hundreds of thousands of tax preparers each serving their own local market and neighborhoods, you have just a handful or maybe just one or two dominant providers who can get most of the market share by providing a higher-quality and/or lower-cost product. And that's certainly what happened with TurboTax. You have a question there? >>: Can we ask questions? >> Erik Brynjolfsson: Let me just say, what we had in mind that Andy and I were going to speak for maybe about, I'll speak for another three or four minutes and then Andy for maybe five or six minutes and then we'll open to questions. That'll probably be best, but if there is some clarification or if I misstate something, feel free to drop it. If you want to talk more generally though, why don't we do that in about five to ten minutes? So where was I? The winner take all markets like for tax preparation software and that leads to a real concentration of the revenues. In the case of tax preparation software, there are about 17 percent fewer human tax preparers now than there were before. Not because taxes are so much easier to prepare than they were before but because technology can now do a lot of those tasks. What I just described in that particular category is a microcosm of what's happening in lots of other categories. Obviously, lots of other kinds of software, but also in music, in media and Marc Andreesen, the adventure capitalist had this great phrase that software is eating the world and by that, I think, he means that these digital technologies are becoming the core of more and more industries. Manufacturing, retailing, finance all sorts of different industries are being affected by it. Someday maybe even my industry, education or increasingly even that industry, and as more of these industries have digital characteristics at their core, they start having the same kind of economics, these winner take all economics or winner take most economics and that has a big affect on the demand for labor. It has a big affect on the income distribution. In particular, there are two sets of winners and one set of losers that are common in these sorts of situations. In the case of tax preparation software, the people who developed that software and who are marketing it and selling it, they've done very well. Some of them are billionaires now, many others are millionaires. Also, another big group of winners is consumers. There's a tremendous amount of bounty. Lots of people are better off because they have access to something that is better, cheaper, faster than what they had before and digital goods by and large are much more accessible to a much broader group of people and lower-cost. But there's also a group that's made worse off and that's the people who previously did that job, the human tax preparers who aren't really adding a lot of value anymore if they are competing with a $39 piece of software and they need to find something else to do, some other way to add value in the economy. And for some of them that's proven very difficult. Many of them had significant training in this area, college educations even, and so that is a group that has been made worse off by this. And that's something that we see happening in lots of other industries as well. So making sense of these patterns of this growing wealth, this growing productivity, the digital bounty that's making many of us better off and the overall society better off is part of the story, but also this significant change in the income distribution and the way the benefits are allocated. And we think that these changes are likely to accelerate in the coming years, if you look at some of the technologies in the pipeline now. I'll turn to Andy here in a minute to describe some of the things we saw as we looked at other sorts of technologies. Ultimately, we think that these kinds of changes in the technology call for a better understanding of what we need to do as individuals, as organizations, as a society that the technology is going to continue to accelerate and the difficulties that we face generally occur because as the technology accelerates, many of us are not keeping up with it. As technology races ahead, many of us individuals and as organizations are being left behind, so once we recognize that issue, we are in a position to maybe do something, take actions and we'll talk about some of the policy recommendations and ideas for what individuals can do to keep up with the technology so we stay more and more in sync. Why don't I pause there and turn it over to my co-author Andy McAfee who was going to talk some more about some of the technologies and the other issues that came up in this book and then we'll open up to questions from all of you and have I hope a robust discussion about this. So Andy? >> Andrew McAfee: Thanks Erik. Like Erik says, we started writing the book because we were confused about some of the things we saw in the statistics, but over the course of writing it and anyone that tells you that they have the goals of their book in mind when they start writing it, is lying to you. Over the course of the book, I kept on hearing this message that I really wanted us to get across as we were writing the book and I kind of wanted to grab the reader by the lapels and just deliver one message which is we ain't seen nothing yet. And all these weird advances that we've been seeing over the past few years with natural language processing and translation and autonomous vehicles and unstructured search, all these things are great. And they are not the crowning achievements of the second machine age, of this digital revolution. They are the warm-up acts and we ain't seen nothing yet. Let me try to make this concrete, and I want to do it -- I'm sorry to inflict this on you. We're going to do a pop quiz. Trust me. You all are going to find it really easy. When I say go please point to a door in this room. Go. Full credit, there are two doors in the room. Please point to where you are in the room. Yeah. And there's this really interesting disjoint. Some folks go like this and some go like this. We'll accept both answers for this. You-all have just solved one of the toughest challenges in robotics. You-all have just solved the problem known as slam; it stands for simultaneous localization and mapping. In other words, what does this room look like? Where are the doors? And where am I in it? What's my localization inside this room? It's a fiendishly difficult problem in robotics. Our evolution has programmed us to be really, really good at it. There was a 100 percent pass rate on our pop quiz. [applause] [laughter]. But you would have an absolute zero percent pass rate with any flavor of robot up until just a very short while ago. It was so bad that we came across a review paper of a discipline written in 2008 that said essentially, man, we're making no progress on slam nor will we at any point in the future. This is just too damn hard a problem to solve. Last year a colleague of ours named John Leonard at MIT did a brilliant demonstration where he solved the slam problem for a room about this size. I am not making this up, by waving a Microsoft Kinect the room. This is a $150 piece of gear. This is consumer electronics. That stuff is now powerful enough to solve the thorniest challenges that we see over and over. In 2004 there was a great book written by a couple of our colleagues saying that people are going to keep driving cars in the future because it's just too hard of a problem for digital stuff to solve. You know the punchline of this joke, right? Over and over again we see this same thing where people were saying, smart informed people in the discipline were saying less than ten years ago ain't gonna happen. Ain't gonna happen. And then these things have started to happen. I'm reminded of a wonderful quote from Hemingway about how a man goes broke. He says it's gradually and then suddenly, and that characterizes this era of technical progress that we have been making incremental progress for a long time on some really thorny challenges and now, boom. And we're just off to the races. And all of the geeks that we talk to -- geek is a term of highest praise, by the way, in the world that Erik and I live in. All of the technology geeks that we talked to say guys, we are just getting warmed up. You ain't seen nothing yet. So to continue on robotics for a while, you know that robots are still expensive, big, slow, heavy, clunky. They can't navigate the world. Have any of us -- I think there are probably some geeky enough people in this room, watched the video of the towel folding robot at Berkeley? Uh-huh, see? They're some geeks in this room. This is a robot that they have that can fold towels, which is actually extremely difficult because they're not, towels are not rigid materials. It takes the robot about ten minutes to fold a towel and you literally have to watch the video and speed it up time because otherwise it's just watching a robot go like [laughter]. It's deadly boring. That robot is going to get a lot quicker in the months and years to come. We're going to have fast towel folding robots before too long. So many of the other things that we think we’re inherently good at, speech, writing, reading, comprehension, taking in a bunch of different sensory channels and navigating a vehicle. Touch, sense, mobility, fine motor skills, we hold out the high ground in a lot of these areas right now, not all of them. We are no longer the best Jeopardy players in the world. I think if we might still be the best medical diagnosticians, not for very much longer. Watson went to medical school as soon as he was done winning on Jeopardy. So this encroachment, this, I don't think we've ever seen anything like this before. There's this digital encroachment into stuff that used to belong to human beings alone and this encroachment is fast, broad, deep and irreversible and making that case has taken about the first third of our book. You'll read that in chapters 1 through 6 of the second machine age. Chapters 7 through 10 or 11 -- I helped write the book. I should probably have the chapters pretty firmly in my head by now. The second third of the book is about the economic consequences of that, and just like Erik said, there are two main consequences. There's good news and then there's some challenging news as well. The good news is more is bounty is more stuff, more volume and variety and higher quality and lower prices and just this ridiculous abundance that technological progress gives it. Honestly, it's the best economic news on the planet today. The second consequence of what we're seeing is a lot less pleasant for everybody. Is spread. It's just this ever increasing divergence into things that we should care a lot about in a society, wealth, income, social mobility, opportunity, stuff that we care a great deal about. There's a lot more divergence in that than there ever has before and if we're anywhere near right, that we ain't seen nothing yet on the technology front, then we ain't seen nothing yet on the economic front as well. The good news is that the bounty will continue. It's my strongest prediction about the future. The other prediction, though, is that absent some interventions, the spread is going to continue to widen as well and there's a lot of reason to think that's fairly bad news for a society as a whole. That's the second chunk of the book. The final chunk of the book is a set of recommendations. What can we do to keep the bounty going while mitigating the spread or dealing with it? We dip our toes into policy world and we offer some recommendations for individuals and then we have a chapter on short term policy recommendations and more radical longer-term policy recommendations. The short term stuff, one of the things that really surprised me and Erik when we were writing and talking about these ideas, is the number of extremely smart people who evidently have never taken a good Econ 101 course. I know that sounds like an insult. I don't mean it that way, but it's amazing. If you grab an Econ 101 textbook and it doesn't matter if you grab one written by a conservative economist or a liberal economist, you grab an Econ 101 textbook and you will learn the same things about effective policies for economic growth. And in particular, what is the role of the government in promoting good healthy economic growth? There is just not a lot of controversy about this stuff. Have fantastic infrastructure, liberalize your immigration, get your educational system right, create a good environment for entrepreneurship, invest in basic research, these are the no-brainers. We're not doing a very good job on all of those things right now, so our short-term chapter is just kind of an Econ 101 refresher. Longer-term, if and when this ridiculous technological progress continues, are more radical recommendations required? I think they might well be, so we go a little bit farther out into strange policy land and talk about things like a basic income or guaranteed income and/or a negative income tax. One of the other main conclusions that Erik and I came to is we really want people to work because when work goes away, individuals suffer and communities suffer to a terrifying extent. The data are just shocking when you read them. So one of our strongest policy goals as we were writing this book was to try to maintain work for people. A very straightforward way to do that is with instead of a positive income tax where for every dollar you make the government takes, you know, say 30 percent, have a negative income tax where people at the entry levels of the workforce work for every dollar they make they actually make $1.30. It's a very strong incentive to continue work and to work, but when we look at our current tax policies one of the oldest insights of economics is tax the stuff you want to see less of and subsidize the stuff you want to see more of. We tax labor right now fairly heavily and that's leading to exactly some of the behaviors you would expect. So we give in to some tax recommendations and get farther out into policy land that way, but the final point that we make in our concluding chapter is that we are heading into, I think it's an overall fantastic future. It's a weird future in a lot of ways. We honestly have not been here before. We have choices to make. Erik and I are both what we call mindful optimists. If we make these choices correctly we're heading into a much, much better world, but don't believe either the utopians who say oh don't worry, you and your silly economic worries. Technological progress will solve all problems for everybody. We do hear that. It's not correct. And also don't believe the fatalists who say there's nothing to be done here. This is some kind of inevitable progression and we're heading into this kind of society where there's a pretty small elite up here and a bunch of fairly miserable people way down here. Don't believe that either. The last line in our book, I'm going to quote and then we'll showed up and open it up for discussion. The last line in our book is technology is not destiny. We shape our destiny. Let me stop there. We would love to hear, our favorite part of all of these is honestly not listening to each other by this point because we've done a lot of that. It's hearing what you all want to talk about, what questions are on your minds. So with that, can we throw it open? Yep? >>: I'm doing a lot of work around business models. Another characteristic that is inherent in visual goods is that they are non-rival excludable and so they makes them public goods. And the weird thing that's going on right now is there's a lot of activity buying back stock instead of investing in capital goods. When I look at the actual business out there, you look at tax preparation, because it's non-rival nonexcludable you end up with three peoples providing it and inevitably somebody says I'm going to take the price to zero and you have the situation where the produce of surplus is basically evaporating and without produce of surplus you can't generate return on investment capital and so you get executives saying I can't generate any pricing power so I'm just going to buy back shares and not invest in the business. And it's kind of frightening because you have a situation like IBM and they are basically doing exactly that. >> Andrew McAfee: I think we would be more worried about the situation you describe if corporate profits were not at an all-time high and business investment in equipment and software were not at an all-time high. >>: If you normalize for buybacks, I don't know if that statement is true. >>: Can someone explain what non-rival nonexcludable means? >> Andrew McAfee: Yes. Chapter 6 of our book [laughter]. It means if I drink this water you can't drink it as well. That's a rival good. This bottle of water is a rival good. If I listen to a song on Spotify, you can listen to the same song on Spotify at the same time. That's a non-rival good. >>: A good example of that is I have a friend who is a software developer and he makes videogames. He used to pay $3500 a seat for some analytic software that allowed him to figure out who was failing various levels and all of that. In three years it's gone from $3500 a seat to free. People can do this. It's easy enough to do with a small number of devs and the prices drop to 0. That's good. Tremendous consumer surplus, but without some producer surplus people say this won't work and so you get a lot of situations. So what I would be interested in is your third book, what happens when… >> Andrew McAfee: Let's not get ahead of ourselves. [laughter]. >>: Surplus disappears because everything is, you know, software eats the world and all the values provided there, but all I need is a couple of devs to jump into name the business and price drops to 0. Let's just take tax preparation… >> Andrew McAfee: I think we've got your point. Thanks. Do you want to jump on that? >> Erik Brynjolfsson: I think that you make a good point about the -- but I want to emphasize the other part of it too, which is a growth in consumer surplus and in addition to having a chapter where we discuss non-rival and nonexcludable goods and some of the properties of them, and I have to say I share Andy's sentiments that right now we're doing all right with profits and by and large with investments. And I can see that scenario sometime in the future, but I just don't see it happening right now. What we also do see though is a tremendous explosion of free goods which mostly are non-rival and nonexcludable. Wikipedia and YouTube and Bing and lots of other services that people can access for essentially 0 cost and that is exactly as you point out, that leads to a huge increase in consumer surplus. There's a lot more value that consumers get without having to pay for it. Is that a good thing? I think that's absolutely a good thing, but ironically, if you look at GDP statistics, it's bad for GDP. The music industry -- Andy was just talking about us sharing music and getting access to music. The music industry has shrunk according to the official numbers, but most of us are listening to more and better music than ever before. This leads to a real problem with the GDP statistics that we aren't measuring the right thing. GDP actually goes the wrong way. When Wikipedia replaces Encyclopedia Britannica, that actually paradoxically has a negative effect on GDP even though it has a positive effect on all of our surplus well-being. By our calculations it's something on the order of $300 billion per year of additional consumer surplus due to all of these free digital goods and that number is going to grow and grow. In addition to the reinvention of some of our organizational systems, we have to reinvent the way we measure the economy because we are not capturing the full bounty. The bounty, the official productivity numbers are going up. What we have discovered in doing this book is that the real benefits are growing even faster than the official numbers because so much of it is not being measured. >> Andrew McAfee: We can make this concrete. Would you raise your hand please if you have and use a GPS system that gives you turn by turn directions when you drive around? Put your hand way in the air, way in the air. Keep your hand way in the air if you had one of these five years ago. All right. What's weird though is that the GPS market is smaller than it was five years ago, because we get this stuff via a free app on our phone as opposed to a device that we pay $200 for. How many of us want to go back to the world of $200 devices? Unless you work for Garmin, that answer is nobody. But it's shrinking; it's literally shrinking GDP. One of the challenges we confront is the fact that our official economic statistics, which guide a lot of decision-making and a lot of policy, they are tracking the actual economy less closely than they used to and less closely year-by-year. >> Erik Brynjolfsson: Right. And the official productivity numbers are driven right off the GDP numbers, so they are equally skewed. Let's take another question here how about over here? >>: I was reading that Princeton released a study comparing Facebook to infectious diseases. [laughter]. They said that basically between 2015 and 2017 they are expecting Facebook to lose about 80 percent of its user base and they are already seeing it dwindle for the teenage sector and stuff like that. Do you think that that's more of a byproduct of the rapid evolution of technology, that people are able to migrate to other locations, other services, platforms? >> Erik Brynjolfsson: This is a characteristic. This is common. It's what Schumpeter called creative destruction. And he and I talk about that a bit in this book as well, that digital goods in addition to having those winner take all properties that we have described, they often also mean that someone else can come along and displace it and just because you have a dominant position at one point doesn't mean you are going to continue to have the dominant position. That could be due to fads or it could be due to better technology and so you get this constant evolution just like Microsoft Excel replaced Lotus 1-2-3 at one point and so you can get this shifting in technology platforms much more rapidly than you did in the predigital era. Question? >>: I haven't heard you touch on it, and maybe it's in your book. If so, let me know. I was listening to a similar talk last October and they were talking about how this acceleration, well in the past when this technology came around, the jobs were lost at the bottom and everything just sort of shifts up. But in this revolution, instead we are seeing right in the middle or even at the top half to or top third, an entire profession go away. Do you talk about that at all? >> Andrew McAfee: Erik's tax law example is the perfect example of that. That is a classic, actually kind of upper-middle-class American job. You need a college education to do that. That's a great middle-class occupation. It is going away because of this digital revolution and colleagues of ours have done fantastic work to highlight this hollowing out the polarization of the economy. Our geek explanation for that is we can't automate a busboy yet. We can't automate a gardener. We can't automate a home health aide. These are all fairly low level professions. We also can't automate a data scientist or a CEO. Those are the high-level. The stuff we can automate is squarely in the middle and it's creeping in both directions. >>: The fundamental ones you mentioned are like medical diagnosis doctors, right, which is considered a high level position today. So the policy challenge really is what do you train people to do? How do you line them up to be ready in five years? You've got a teenage son or daughter, what do you tell them to study? We used to know. Just look at the strata. >> Erik Brynjolfsson: That's exactly. That's a good point. >> Andrew McAfee: Yes we do touch on it. >> Erik Brynjolfsson: And we spend a whole chapter talking about this. It's still true that the routine information processing tasks have been the hardest hit. If you look statistically, those are the ones where computers have made the biggest inroads. But they are beginning to creep into these other areas. That said, tasks that involve creativity, also tasks that involve interpersonal relations, interacting with other people and understanding their feelings, those are also things that machines aren't particularly good at. Those are two areas where humans have an advantage these days. But as you suggest with your question, the pace of change is quickening and having, you can't expect to have a 40 year run with one particular profession or one particular task. So in addition to learning those skills, you also need to learn to adjust and constantly update and change your skills. To put this in a little bit of context, it's important to understand computers have always been, not computers. Technology has always been destroying jobs and it's always been creating jobs, but the pace at which it's happening is happening a lot more quickly, so our need to adjust and adapt to that is happening a lot faster as well. >> Andrew McAfee: It's a really, really tough question to answer because the answer that we would have given with confidence ten years ago is basically hone your pattern recognition skills. That's a joke now. Computers are much better than us at most flavors of pattern recognition. That said, Erik and I in all of our work never came across a creative computer, an innovative one an entrepreneurial one, so this idea of ideation, of coming up with something new still seems to be uniquely human territory. We've ever seen a computer that could motivate a team to go do something, that was a really good negotiator, so these kinds of deep interpersonal skills are really, really valuable. But I think most fundamentally, if you've got kids, teach them to love to learn because they're going to have to learn and shift multiple times in this world that we're heading into. >> Erik Brynjolfsson: Another question? >>: Maybe a bit of a naïve question, but when you think of the jobs that were lost by the tax preparing people weren't there more jobs created for people who write the code for the tax preparation software and thanks to our IRS who change rules every year so there are always going to be need for that? It's like energy, right? It changes from one form to another. [indiscernible] [multiple speakers]. So what you are saying is the rate of destruction of jobs, the rate of creation of jobs there is a massive disparity. Have you done any analysis that says that the percentage in the first machine age and the percentage now, what is that percentage? >> Andrew McAfee: Your question actually is not a naïve question. It's a really, really deep, really good question. For 200 years people, allegedly smart people have been saying technological unemployment is right around the corner. Have you heard of a group called the Luddites? Twenty years ago these were British fairly high skilled workers in the textile factories who started smashing the automated looms because they were afraid of unemployment. All of these people predicting technological unemployment have basically been wrong. For 200 years we've had huge amounts of technological progress and we've had very close to full employment in the rich world. That's because exactly like Erik said earlier, it's easy to emphasize the destruction and a lot harder to emphasize the creation that takes place. The creation keeps pleasantly surprising us. The question that we go back and forth on all the time, and we don't have the final answer on it at all, is is this time finally different? My gut tells me it is. We've got to stay tuned. >> Erik Brynjolfsson: And specifically, the data say that for the past ten or 15 years it has been different. We have had a divergence in job creation from what it was historically. Even before the great recession, and it's been especially sharp in specifically routine information processing tasks. We can look at particular categories like tax preparation software and there are a lot fewer people doing that than there used to be even accounting for the relative handful that are involved in the software companies that are producing that. But colleagues of ours at MIT and elsewhere have looked at the skill content of all the occupations in the United States and what kinds of skills are required of them. You can rank all of the occupations by how much routine information processing is required in them, like tax preparation, bookkeepers, clerks, travel agents. The more routine information processing in the job, the more it has been in decline in the total demand, the total number of jobs and the wages in those categories. So that's pretty strong evidence that they're being hit quite hard. Can people shift into other categories? Absolutely, if they can, they should. How long does that take? Five years, it could take ten years, but of course, by then the technology has shifted again as we just heard from Andy. There's a whole new set of technologies in the pipeline that are affecting other parts of the spectrum, so there's going to be this constant race, this constant catch-up and it becomes more and more difficult to keep up with the technology unless you accelerate your re-skilling and the reinvention of work. >>: How many web designers are there out there? It seems like that's a huge growth area. >> Erik Brynjolfsson: It is a growth area, and actually there are areas that are growing. There are areas that are shrinking. Net, net, there's been more destruction than creation in the past ten to 15 years. >> Andrew McAfee: But we've been through, to your point, we've been through long transitions before when there was sustained pretty high unemployment and we reached a new equilibrium of pretty close to full unemployment. Again, the historical pattern should call me down. I just don't know how much history is repeating itself this time when cars drive themselves and my phone talks back to me and computers are the best diagnosticians in the world, the shrinking set of stuff that we need people for winds me up a little bit. >> Erik Brynjolfsson: I think there was a question in the back, how about all the way back there? >>: A lot of this talk seems to be on U.S. focus. What about the 5 billion people that live in the rest of the world? >> Erik Brynjolfsson: I think that's a great question so let me break those 5 billion people into two broad groups. Let's look at other developed countries and then let's look at more developing countries. Other developed countries there is an organization of economic cooperation and development that gathers data from those relatively wealthy countries, Japan, Western Europe. If you look at them you see remarkably similar trends in all of them on some of these key metrics. For instance, inequality has grown in 18 of the 22 countries that are surveyed. It's grown in Sweden, in Germany, in Japan. Two it stayed about level and two it reduced a bit and that was Greece and one other one and I don't know how representative they are of economies that are on the right track these days [laughter]. It seems to be a worldwide trend. It's not, they started at different levels of inequality, but they are all having a directionally similar impact. That suggests to us that there is an underlying force that's affecting all of these countries. It's not just politics or particular institutions or culture in one country or another. It's something more pervasive and worldwide and fundamental. Let's look at the developing countries. We've looked at what's happening to some of the employment trends there. Take China, we spent some time visiting there. Many people talk about globalization and the movement of jobs from the United States to China and certainly globalization and technology are probably the two big forces right now. But if you look more closely at what's happening in manufacturing, for instance, what you see is -- what do you think is happening actually to the amount of manufacturing jobs in China as that industry grows massively over there? It turns out that the number of manufacturing jobs in China is dropping rapidly. There's 20 million fewer manufacturing jobs today in China than there were in the late 1990s. There are fewer manufacturing jobs in the United States as well. So it's not so much that the jobs are going from the United States over to China. It's going from both the United States and China to robots. People like Terry Gou who runs Foxconn is talking about putting in place one million robots to replace half of his workforce assembling iPhones and other phones over there in China. Our view is, yes, globalization is another important force and these other countries are being affected, but ultimately off shoring is just a way station on the road to automation, and in many ways those low-wage workers who are doing relatively routine tasks are even more in the bull's-eye of some of these powerful technological forces than the relatively smaller number of workers who are still doing some of those tasks in the United States and other developing countries. >> Andrew McAfee: The research is really adding up and it's becoming clear that both of these economic phenomenon that we talked about, the bounty, is not just a U.S. phenomenon; it's global as well. That's the good news, but spread is going up in almost every country around the world as well. The examples we've given so far are kind of U.S. world examples. The phenomena are pretty much global examples. >> Erik Brynjolfsson: But of course, overall income, and I want to stress the bounty part too. I mean, let's not lose sight of that. The overall wealth in not just the United States but worldwide has gone up tremendously. How about the far corner over here? >>: [indiscernible] do you find a similar correlation [indiscernible] jobs going down or is it [indiscernible] >> Andrew McAfee: Similar correlation with what? Sorry I didn't hear the first part of the question. >>: Jobs creation and food production. >> Andrew McAfee: Food production. Okay. Agriculture is a really interesting example because that was really the big industry that was being affected in the first machine age. At one point 90 percent of Americans worked in agriculture worldwide for that matter. Today, it's less than 2 percent. Now all of those people didn't become unemployed. They got reemployed into new industries over the course of a couple of hundred years, this is what happened. Part of the reason they became reemployed was because they learned new skills. We talked about that. Some people have said the best idea America ever had was mass compulsory education that helped provide a set of skills for a workforce that previously had been involved in agriculture and then became involved in industry and services. There's also a big role for entrepreneurs. Who invented those new industries that those people were involved in? It wasn't MIT professors. It wasn't government policymakers. It was people like Henry Ford and Bill Gates and others who helped come up with entirely new industries that people couldn't even have conceived of at that time and they had need for a lot of workers in those industries and they reemployed those. So that combination of some enlightened government policy that helped re-skill people and some real powerful entrepreneurship that helped invent some of those new industries ultimately reemployed tens of millions of people from agriculture into other sectors. We think that's a bit of a model and hopeful for what can happen going forward. Now of course, it's not so simple that we can simply mimic what was done before and say okay we are going to have compulsory, I don't know, college education. It's not nearly as simple as that. We have to reinvent education more fundamentally to the kinds of creativity and interpersonal skills and a whole other set of things that we need to learn and we need to invent a new set of industries that apparently we are not doing fast enough. That's what the data suggest. In fact, the amount of job creation from new startups in the past decade has been smaller than it was in the ‘90s or ‘80s. You may hear some buzz about lots of entrepreneurship, but it is not actually happening as fast as it used to. So we need to make progress on both of those, but I think the example of agriculture and food production is one that we can learn a lot from. >>: So you said that you were going to have to face one of those starry eyed optimists in the beginning and I appreciate your lecture and I think you are exactly right and I wouldn't disagree with anything, but isn't it possible to take that back set and look at it a different way? I've done those jobs that have been eliminated. Frankly, I hope nobody ever has to do that. [indiscernible] it really sucks. But there is a breakthrough that you haven't talked about which is the smart guys over at MIT could figure out how to magnetic finding for fusion, we could have coming I mean the fancy term for the charitable good we could literally have unlimited power and then unemployment doesn't really mean anything because there's enough of those physical goods forever. You could call it the Star Trek model. Why do you think that we won't get to the Star Trek model acknowledging that change is going to go forward? Why don't we get that tomorrow when somebody hey look. A pitcher of seawater produces enough power for everybody in China? Andrew McAfee: Right. We're already getting close to that, closer and closer to that Star Trek model in a lot of ways. Most of us and more of us all the time are being freed up from material want, from deprivation. To me that's heading towards that Star Trek universe where you push the button and your meal comes out or whatever. We are getting there. Honestly, that's not enough for a healthy society. We came across a fantastic quote from Voltaire of all people when we were writing the book. He said work saves a man from three great evils, boredom, vice and need. Of those three, the need is the easiest one to take care of. My question is in this sci-fi future that you're articulating, what is the meaning for life look like? What does a good life look like? What is a fulfilling life? What does a healthy community look like? We have come nowhere near solving that problem. When you look at communities now where work has gone away, in America anyway, you don't see starvation. You don't see malnourished children. You don't see these terrible material wants. You see children growing up outside of two-parent homes. You see divorce rates skyrocketing. You see crime skyrocketing. You see people not voting anymore or participating in civic life. You see this really ugly breakdown in the community and that's the problem that I find myself, I think we find ourselves a lot more worried about than the how are we going to provide for the material needs of the people. >> Erik Brynjolfsson: Andy, you're usually -- sometimes we have this conversation. You are usually a little bit more optimistic when we talk a little bit far out into the future and I think that that's a set of concerns that we have in the short run. But I think that that Star Trek future of the far future is potentially one that could be very, very appealing, one of a great deal of abundance and I think there's a realistic possibility that within say in the next 20 years we will for the first time in history eliminate severe poverty on the planet. That's the trend that we are on. I don't think that it's going to happen tomorrow or the next year because these things do take time and even if we had fusion reactors it wouldn't be sufficient because energy is only one input. There's a lot of other inputs that we need to deal with. But we are on the path towards having a lot more abundance than we used to have. It's going to be a very rocky transition for several decades and we're not quite sure what that Star Trek future will look like and how people will get a sense of fulfillment and meaning in their lives. It's something we talked a little bit about in the book towards the later part of the book, but much of the book is focused on the next decade and the decade after that where we see a great deal of disruption because unfortunately, most of us aren't going to have replicators in our homes any time in the next few years. Although, my friend Neil Gershonfeld, [phonetic] our friend, in the media lab, when I went around at the dinner about five days ago that's how he introduced himself. He said he was working on a Star Trek replicator, so good luck to him. >>: I wonder if the research that you were talking about [indiscernible] >> Andrew McAfee: A little bit louder. >>: Are unions going to play out in the future? >> Andrew McAfee: Unions? >>: Yeah. You know, will they preserve jobs or will the unions just go away because the blue collar workers go away? >> Erik Brynjolfsson: To have an effective union, the most important thing you have to have is some bargaining power and bargaining power comes from having something that the other party needs and that they can't do without. What's happening increasingly is the technology is weakening the bargaining power of the working man to use that terminology. >> Andrew McAfee: As is globalization. >> Erik Brynjolfsson: As is globalization, and that is one of the main reasons I think that unions are having a lot less power than they used to. I think if you just sort of tinker with the union laws and all of that, that's sort of like rearranging the deck chairs. It's not really dealing with the more fundamental transformative issues that are happening in the economy right now which in our view half to do with part globalization, but especially the underlying technological trends. >>: We've talked a lot about job displacement, sort of one for one, car industry workers go away because of technology. Did you look at any of the collateral effect, you know, safer cars and the effect on insurance agents because you need less adjusters and that industry falls apart because of those kinds of things? >> Andrew McAfee: It's a super complicated situation, but I want to pickup on one aspect of what you said. When we talk about bounty, and the reason Erik and I keep talking about the bounty part of the economic implications, is it's too easy to overlook and it's too easy to get gloomy about some of the really tough statistics we see about jobs and wages in the middleclass. It's important to spend time on those, but also important not to lose sight of the good news part of this as well. You mentioned safer cars. You know that we have fewer accidents in America than we have had at any time since I can't remember if it's the ‘60s or the ‘70s? Not per mile driven, overall total in America road fatalities are dropping every year even though there are a lot more of us and a lot more cars and we are driving faster, that's a safety story. It's a technology story. It's a quality story it's all about this bounty that the world is, that our world is creating for us. We think it's really important not to lose sight of that good news. >> Erik Brynjolfsson: And much of that is unmeasured in the official statistics. One little amazing fact that is in the book that you will see, is that if you measure the value of the increased safety and quality of life here and the longevity that we've had over the past 50 years, that any reasonable economic estimate of that value is greater than the value of all of the other innovations put together. That doesn't show up in the GDP statistics, that improved quality of life. So that is just one… >> Andrew McAfee: GDP can't count someone who doesn't die in a car accident even though they were hit very hard. >> Erik Brynjolfsson: Those effects are real and they are enormous and they reflect the fact that we aren't measuring our economy properly. How about over here? >>: I was going to follow up on the guy that mentioned about Star Trek. Do you think that our sense of challenge is waning? So in jeopardy, I mean, Watson could go out and beat anyone in the world. In other things, like machine learning, are machines going to become innovative, perform [indiscernible] over sets of data and come up with new experiments to manufacture something, put it out there and see who consumes it and if it doesn't try something else? >> Andrew McAfee: That's a philosophical. That's not what I see it all. Maybe I'm hanging around with a non-representative group of people. I hang around with the folks here at Microsoft and Silicon Valley, MIT and I've never seen people as excited and as creative before. In fact, I would make the argument as physical deprivation gets taken care of for a bigger and bigger chunk of the planet and as more and more of them get connected to the global internet, we're going to have a lot more people being part of the creative community. Not only can billions of brains, human brains now tap into the world's knowledge, something that was never possible before, but perhaps more importantly, billions of people can contribute to the world's knowledge in that way. We have a whole chapter on something we call the combinatorial innovation which is the idea that most innovations don't just spring out of nowhere. They come from combining and recombining pre-existing ideas and the more people you have contributing to that stock of ideas, it doesn't use up innovations for other people. Instead, it creates building blocks for other people to add to. And I think that as we get more innovation, including some of the ones you mentioned, it only provides more fodder for the next generation of, in some cases teenagers to build on the shoulders of giants in a way that they couldn't have before. >> Erik Brynjolfsson: Keep in mind, we still play Jeopardy. It's still a very popular quiz show every night even though a computer is better. There's still a lot of chess players in the world. None of them can beat anywhere near the best computer anymore. So I don't think that digital progress necessarily stomps on our human aspirations and makes us go slink away. Luckily, that's not what happens. >> Andrew McAfee: It's going to be the Olympics, right in a little while and people are going to be racing 100 meters. >> Erik Brynjolfsson: In four more years we'll have a speed skating robot that's faster than any human and we'll still have speed skating at the Olympics for some time soon. >> Andrew McAfee: How about right up front here? >> Erik Brynjolfsson: That'll be a cool robot, won't it? >>: I want to know your thoughts about the people who get displaced from their jobs again. For example, you gave [indiscernible] office [indiscernible]. It takes years of training to get to where they are and it also has become an area of entry and that's how they get paid so much. >> Andrew McAfee: The reason they did that expensive CPA is to go get a job as a tax preparer and have a good long career. >>: So like after [indiscernible] software they are not prepared to shift to other high-paying career right away. So that's why we have the high unemployment because [indiscernible] taken their job. So how do you recommend to make this transition? >> Amy Draves: To make it easier could you repeat the question? >> Andrew McAfee: Sure. People find that the investment that they made in what is commonly called human capital in these skills becomes more rapidly obsolete because of technology and these other forces. What should we recommend that we do? How can we respond to that? And I think that the answer, to build on what we said earlier you have to build a world where you can adjust much, much more rapidly and that's true at the individual level. We're all going to have to work harder, run harder to keep up and learn more skills. Maybe digital technologies can help us learn those skills faster than we did before. As a society we're going to have to smooth that transition more easily. Entrepreneurs are going to have to be more creative about finding new ways of redeploying people and using those skills. I think that they haven't been doing it fast enough. We love entrepreneurs, not because we think everyone who becomes an employee is then going to become an entrepreneur. It's not like that's the job that they end up in, but it's because in our society entrepreneurs are the people who are in charge of coming up with the new industries and the new jobs. They're charged with the creative task of figuring out what that person who you just described might be able to do next along with the person himself. So it's really a joint effort between government society, the individual and the entrepreneurs to work on that and it has to happen faster than it did in the past. People say we have lots of entrepreneurs and we need more. People say we are investing in education and re-skilling and we need to do that faster. So all of those things are going to have to accelerate. >>: [indiscernible] workers that they need. >> Andrew McAfee: You asked a really good question. I find it a really tough question. What do you say to a CPA who just doesn't have a job anymore? I find that an almost impossible conversation to have. Pick yourself up. Learn some new skills. That's how capitalism works? Yuck. I mean that's the terrible thing to say to somebody. So instead we try to concentrate the conversation elsewhere and say to policymakers and people involved in the educational system, look. You have got to create a very fertile environment, not an environment that makes sure that that person can never be laid off. That's a terrible idea. We can't freeze the economy, but we need to create an environment where when that person loses a job, that there's a good environment for them to move into. There are jobs available. There's actually an educational system that will give them skills because when we talk to business leaders we hear over and over what you just said, which is I can't find the skills that I need. I've got hiring wants. I can't find them from the entry level all the way up to the very top. Our educational system is not doing a great job in this very tough time. I want to try to shift to your question, because it's such a difficult one. Instead of talking directly to that laid off person, let's talk to the other institutions around and say what can you do to help? >> Andrew McAfee: Or in addition, I think everybody's going to have to take some responsibility. I'm looking at the clock here and I think we want to leave some time for signing books, so why don't we just take one last question and that we can talk in some one-on-one conversations. >> Erik Brynjolfsson: I think our host is going to take host prerogative and asked the last question, only if it's a really good one though. >> Amy Draves: It's an online question. There's been, it's been blowing up here. There is a question around ethics was discussed as Google deep mines [indiscernible] without and then also… >> Erik Brynjolfsson: Pick one question and hit us with it. >> Amy Draves: Choose what you like. >> Andrew McAfee: Okay. Go-ahead. Go-ahead. >> Amy Draves: He wants to know about licensing and regulation and if that's going to assist. >> Andrew McAfee: Let's talk about licensing and regulation little bit. A lot of us, stupid question time. How many of us are familiar with Uber? How many of us have Uuber on our… Okay. Uber is a great example there because they are finding themselves fighting all kinds of regulators and when I look at that fight it seems to me that what we're seeing is incumbents and entrenched interests trying very, very hard to protect themselves and make sure nothing changes. My favorite example of that is, you know in France now, Uber has to bake in a 15 minute delay before it can send a car to you, because the taxi drivers in France in Paris were so upset about that. That's the wrong kind of regulation, an extremely stupid kind of regulation. You guys know who Tim O'Reilly is? He is one of the gurus of the tech industry. He is a buddy of ours and he has a beautiful way to phrase it. He says look. With our regulations and our policy decisions and our licensing and all of these other requirements that we have, you have a choice. You can protect the past from the future. That's the choice that France is making with these taxidrivers. Or you can detect the future from the past; that's the choice we really should be making because the future that we are heading into is a better place than the one that we've got today. >> Amy Draves: Thank you so much for coming. >> Andrew McAfee: All right. Thank you all for showing up.