Yuma 4×4

Media and Communications

LinkedIn Speaker Series: Erik Brynjolfsson, Andrew McAfee, and Reid Hoffman

LinkedIn Speaker Series:  Erik Brynjolfsson, Andrew McAfee, and Reid Hoffman

♪ >> >>Ladies and gentlemen please welcome Alan Bloom.>>Good morning everybody, my name is Alan blue, I’m one of the cofounders here and I’m also vice president of product here. Thank you all for joining us for this edition of the Lincoln SpeakerSeries. We put these on a regular basis to try to bring some of the smartest, most thoughtful people here to linked in. People will talk about the stuff that matters to us as a company, but also to the vision and mission we have to the world. And for those of you watching online, and here, if you want to go to speakers. LinkedIn.com, you can check all the old ones and see what is coming up on our SpeakerSeries. As you all know the question around the future of work is a big deal for us here at LinkedIn. Not only do we concern ourselves with where we are today, but because we are members first because we want to make each individual successful and provide each individual with economic opportunity. The future is both easy to see it harder to see than it ever has been. Technology is changing things very rapidly. There are beginning to emerge a set of voices around the world who are beginning to think about these topics. And we are very pleased today to have two of those voices, two of the people who originally started talking about what it means when machines begin to act and they will talk more about it, begin to act not just in lieu of muscles but in lieu of our brains. What does that mean for us on a go forward basis? It’s the author of two books, both from MIT, and people who have had the pleasure of working with and speaking with on many occasions, Eric and Andrew McAfee. Please,. And my cofounder Reid Hoffman is going to engage them in conversation. Enjoy.>>So the fortunate thing is we are at least a few years off from machines replacing our ability in these panel conversations. Just so everyone knows I’ve been a fan of Andrew and Eric’s for a long time.>>Just so you all know our main strategy is to talk to read whenever we can and in secret transcribe what he says. And put covers on and try to sell copies.>>Let’s start with why did you write this book? What is your target? What are you trying to do as we know the three transformations that are happening but why did you write this book?>>I think we are in the early stages of a huge revolution. Andy and I have been looking at the way technology has been transforming, and all of you in this room have been working to lead a lot of this transformation. But what you may not be as obvious he was away the rest of the world, the economy is being changed. We’re looking at the companies that are being disrupted and transformed. We look ahead and we see the next 10 years potentially being the best 10 years in human history or some of the worst because there is just some huge challenges. The gap that we seek to address is not that the technology is moving fast, it’s that the rest of us are not keeping up with it. Economics is way behind in human institutions. And so in her previous book we talked a lot about the disruption that is coming, the bounty that’s been created but also the inequality that’s being created. And this we look at these three great rebalancing trends that read says would be a better word between mind and machine, product and platform, and the core on the crowd. And this triple evolution is going to change the way a lot of businesses need to be run, the way a lot of people re-skill themselves. And the way we think about society more broadly.>>We were on the road and talked about a lot and we kept noticing the were having the same hallway conversation over and over. Which is you go on stage and we would be accosted in the hallway by somebody saying I run a medium size to big-company industry x, I believe the story that you are telling, what do I do now, everything that the world or my business differently. It was a great question and Eric and I are both business school guys and we thought it was incumbent on us to try to povide answers to that question. The problem was we didn’t know the answer to that question at that time. It was scary but also a great opportunity because the two of us get to go into a room with a white board and try to figure stuff out. Try to talk to some of our favorite geek friends out there and try to answer this question.>>We are really looking forward to the Q&A and discussion from raid and all of you so we can hear more questions about things we should be thinking about getting to work on. It’s another one of our secret formulas to try to do better work is to hear from some of the best and the brightest about what you are seeing.>>When we went out in the world we knew that we had to talk about some number of things that were happening. One is not enough. I got really good advice earlier from somebody that said if you ever find yourself on stage saying and six fully, you are in a lot of trouble. Success to money. It has to be a finite number of things that people get their minds around. That’s for this idea of three rebalancing’s and shavings came from? The good news is that we learn that with there are three separate bodies of Nobel prize-winning economic work that have a huge amount to say about the stuff that is happening today. The economic backbone of the book is one of the things that I am proudest of.>>I think I am a word pennant today.>>This panel is secrets revealed.>>What are some of the things as traditional industries and companies look at the transformation happening with technology to have this kind of massive quick changing within the world of software. They have a need for different skill bases, I need for different strategy for the business, a different approach to data. There is obviously lots of things to do and don’t do. One of the pieces of advice is not everyone should try and become an AI first company next her because I would be a mistake. Even though AI will transform all industries. What are some of the cheat sheet on the do’s and don’ts for traditional industry folks?>>They should be a lot more data-driven. Not everyone is going to become LinkedIn or Google or whatever but in terms of resetting the dial, our experience is most companies and organizations are not nearly data intensive enough. Not nearly technology intensive enough. And too many decisions to be made by human judgment. We spent time talking to Nobel prize-winning people like- we are just astonished by the number of human foibles and mistakes that happen over and over.>>There are about 175 entries.>>We will run through all of them today.>>And soap the first one is people are way over confident in their own judgment. We look at >>Raise your hand if you have below average judgment?>>We did a bunch of studies ourselves and looked at how if you did more data-driven decision-making you could do a lot better, and I briefly pointed at her own decision-making. We tenure people based on this committee that sits around and makes her human judgments, and along with some grad students and in operations research we did analysis, we call it Moneyball for professors where he took a bunch of data about citations and publications and other kinds of data and you did some more sophisticated analytics on those, you could predict who should be getting tenure and who didn’t. You compare the list that the model created with the list of people who actually got created, they overlapped about 70 or 75% of the time. Most interesting is the ones that he disagreed on, the model, those people do substantially better in terms of having high impact research. The research got more citations, had more impact, and the subsequent years. If you want to have tenure people who did great research she would be better off going with the model. Our book is a little more nuanced.>>This is a debate Eric and I had because I am above the school that you should throw up the human decision making. Eric has more fondness for our wetware.>>I kept on citing actual evidence.>>You generated a model that showed that I might’ve been convinced. But that is part of the fun of writing a paper we have these creative tension and debates. I remember these discussions over lunch. We sat down and I think we came to sort of a meeting of the minds that using a lot more data and models would be very beneficial. I still don’t think that we can just go all the way that way for most problems, but many of them would be way better off turning the dial further in that direction.>>The single most failure I see among enterprises that we’ve worked with is the managers still think a big part of their job is to be the gatekeeper of ideas, to pass judgment on the ideas that might make it out to the world or not. They do that based on their gut, their experience, their expertise. Our favorite business acronym is hippo which stands for highest-paid persons opinion. It’s how most decisions get made in most places. The thing that I think I am most impressed by is how people that run geeky companies are trying to get out of the hippo business and try to create a culture that lets the evidence speak that has an honest conversation and tries to be a lot less hippo driven.>>When I was doing some of the conversations behind the book that I will be publishing this year, one of the interesting things I discovered was the way that basals manages. things I discovered was the way that Bezos manages. But if you have data you can overwhelm my opinion. Part of his attempts to make that if I disagree with you bring data. One of the more interesting examples of that was the Amazon feature of asking a customer to write an almost like Wikipedia like information about the products. Basals thought that was a terrible idea. So they papered it. So they papered it.>>That is great advice actually and that is what we talked about the book as well. This culture of experimentation and maybe testing. Scientists have been doing this for a long time. But businesses outside of LinkedIn and Silicon Valley, it is fairly rare at Amazon to do those tests. But because of the digital infrastructure we have now instead of debating in a room and people having my opinion versus yours, you said let’s run an experiment. If you have a website you guys can run hundreds of A.B. Tests each week.>>We interviewed the CEO of audacity and he said they were employing people to review code. One of the engineers said we have a lens out there, why don’t we just contact them to redo other people’s code. And he says go try it, go see if tpping into the crowd that we will work. He didn’t even attempt to exercise his own it seems like a good idea to me kind of.>>But that is rare. People like him and Jeff basals are the exception, not the rule. We spend most of our time talking to the CEOs, and it is a real ctural shift from being a hippo to being I will step back and let the data speak. People like him and Jeff I will step back and let the data speak.>>Have you been thinking about creating or have you already created >>Several. There are two courses while three that are sort of touching on this. One is a PhD course. There is an MBA course, economics of information. And then there’s a third, and analytics lab which is all about project-based analytics projects. We team up with a bunch of different companies from around the world, they give us a very large data set. In September the students make teams and they work with these 20 million item data sets and they spend the next three months analyzing them and in December we spend the whole day having the students present back to the companies what they discovered. Every single one last year they significantly outperformed with the company was able to do with that data and came up with new insights.>>I’d taught at Harvard business school for a decade before I came back to Sloan. All I was encouraging students to do on that decade was being credibly fond of their own judgment after reading eight page case about a complex business situation.>>This is what is interesting as you would think the classic MBA methodology popularized by Harvard is the case study. You think you would entirely change that methodology based on the advice you are given which strikes me as extremely sound.>>Working on it.>>If you were to look at the Sloan school, we are much more project-based data driven and moving away from the case style. Now much more of the projects we call them action learning for we are working with data and putting in place some of these principles in real-world situations.>>LinkedIn is also trying to help solve this problem because roughly speaking whether it is information from the newsfeed, whether it is LinkedIn Learning, or Lynda.com, it is how we have people continually adapt to new skill sets to be successful in any industry? As a primary mission in this intersection between the stuff that we are doing in the stuff you are kind of highlighting in addition to saying the experimental, oriented towards data, use judgment but factor data into your judgment. Don’t think it is my intuition as much as my judgment of data is important. Is there anything you think that LinkedIn could be doing that you would think of? This is an on the spot question so it may be but that is something that we also treat as a very fundamental issue.>>You guys are best positioned of anybody to address this kind of problem. You have data on all the jobs people are doing, how they move, what their career ladders are, what their skill gaps are. I think you could be doing a lot more but you are very well positioned to understand these skill gaps. In our books we talk a lot about the way that there is a growing mismatch between the skills that are needed, the human capital versus what the technology the people are doing. The way we first met Reed was at Oxford University some years ago. We wrote a pamphlet type book called race against the machine and we had a debate. It was very much on this topic of how >>The single smartest decision we made was to let Reed be the anchor person for our team.>>We were on his team and we won. It was very clear to us that this is a huge challenge and challenge for society. The first part of the challenge was to understand what exactly are the skills mismatched? Where is it that we don’t have enough people do and how do you get that data? I’m looking at a roomful of data. I see you guys, I’m glad you’re leading the challenge. Allen that is a really big part of our solution to that problem. You guys are well-positioned. I think you could be doing more but I’m glad you’re doing as much as you are.>>There’s another aspect to this which is that a company like LinkedIn is in a good position to help us solve an important really fundamental puzzle which is what are the characteristics of a lifelong learner. Had a somebody managed to do that throughout their career. The dirty secret about our industry of educating people is we have a huge number of very deeply held opinions backed up by an incredibly small amount of data. And flipping that Iran becomes an important thing to do. I was debating with our mutual friend Peter Teal. I disagree with Peter on almost everything. The one I was debating with our mutual friend Peter The one thing the one thing he said that I agreed with was everybody says they are teaching critical thinking, most people are not. How you actually go about building that skill and having them be or execute a successful multistage career we don’t know about that, you guys do.>>Let’s switch a little bit to society. One of the things that I have watched with some consternation over the last year is this growing thing that has called tech wash which is roughly speaking in a overly simplistic diagnosis. You have a growing political storm around technology, even if a little bit less popular. The growing lashes a combination of Republicans Democrats going wait a minute maybe this was a problem for the election? Maybe there’s a large company and large companies are not normally our friends. And so they are coming together. For me, this strikes me as a highly dangerous thing because while I think there are challenges with technology, technologies how you improve it. How do we get there. What is your kind of thoughts on tech last? What you think society or politician shins should do? What you think tech companies should be doing?>>Eric is a good three part.>>The first digit is that tech is doing amazing things for society. It is one of the best things that could change so many things.>>Tech progress is the only free lunch economists believe in is the old joke.>>That said, there is a tech wash. It’s not just made up opinions, there are some real challenges and I think tech companies have been remiss in not getting in front of these challenges. I will give you a seven part lists.>>We are just avoiding six.>>There is an economic challenge that so many people are being left behind. Median income is stagnating by most formal measures. There is a related thing that big companies are getting winner take all markets because of networks. Reed has written a lot about the stock that leads to a concentration of power. A third thing is that there is a cyber risk that is more or more of our infrastructure is digital we become much more vulnerable. Whether it is hacking or tax. A fourth is that there are a lot of developing countries that have had is a plan to do a lot of low-wage labor that allows him to catch up and has worked for some companies. That bridges being taken away right now because machines can do a lot of that kind of labor. These are for economic challenges. There are also three other challenges I will touch quickly. One is privacy. Our phones, the Internet of things, we are constantly broadcasting a stream about us. There are cameras and more cities that have face recognition. We are in any world where we don’t have the kind of privacy we once had. There are issues with algorithmic biases you have machines like more of these decisions. They typically learn from human decisions. Humans make flawed decisions. That is a risk if you embed them into the machines. I think we can do better than we did before but if you just blindly have a black box that mimics what we are doing, then you are going to do what we were doing. Last but not least, there is this whole issue around FAKE NEWS and amplifying the echo chamber that has gotten a lot of attention. But it is taking real news, bad thing being amplified 10,000 fold and trying to get you and him fight to get angry at each other and that kind of cyber bullying could be ample fight as well. There are at least seven challenges that are real and legitimate that tech companies need to take the lead on. If they don’t it’ll be the Democratic Party, the Republic and party, the media lots of other people who set the agenda and not in ways that are friendly to tech companies but perhaps more fundamentally the right kinds of solutions for society.>>This is one of the very few areas in politics today whether it is some flavor of bipartisan consensus, that the big tech companies are bad. Eric gave seven reasons why they might be bad. This is big scary bad cloud floating around. I come across a lot of hippo thinking people think I am vaguely uncomfortable about Amazon Apple Facebook Google Microsoft and therefore something must be done. That is deeply shoddy reasoning. And I will do Eric’s test for them. He’s got a nice way to think about this. Let’s look at the economic impact of this in three ways. How our competitors are doing. You want to make the next social network. There are a lot of people saying they are doing free R and D for the giant tech companies so there might be a chilling effect on competition. What is a state of innovation these days. One of the things he worry about with monopolists is they tend to stop innovating. There is a store in Seattle where you put down the bag and walk out the door and you are no longer shoplifting. Those companies are among the world’s largest vendors are R and D. The third leg is how are all of us benefiting out there? We are getting a cornucopia of free stuff and the biggest knock on Amazon is at the prices are too low. It is really hard to see how consumers are being harmed by this. If you are batting 667, you should keep going to the plate.>>I want to write back to one thing in the book, I realized it was something we didn’t cover that would be useful. The economist advice process. Part of what your packaging is there actually a bunch of good economics work that actually should be wrapped into the business practice. What is the quick summary of that?>>Act asking economist for a quick summary and see how that goes.>>The short answer is we do book. There is some really fundamental economic work on how you do decision-making, develop platforms and leverage them, how you leverage the core of the crowd. I will pick the platform one. Platforms can be thought of a lot of different ways but one thing we found useful is this concept of two-sided networks. Some of the interesting things is one-sided networks, on one-sided network is a phone or a fax machine or a what’s up. The more other people who are using the same thing, the better off you are. Is not value unless other people are. When there are two different products, two different groups of people and yet they benefit each other. September there is an app the drivers use and there is a different related up the customers used in the more people there using the Cooper app on one side the better it is for the people on the other side. There are lots of other places where you have these two side networks thought it turns out that there are times when it really does pay to give stuff away for free to build the other network. You may even want to give it for less than free. The credit cards are two-sided networks. Originally they used to charge the merchants, the customers for the credit cards. Now how many people here get paid to use your credit card? Most of you probably do. If you get frequent flyers they are paying you if you are smart to use the credit card. Is providing a service but it turned out the elasticity of demand is such that it is better to subsidize the price to free or even less than free, get more of you using a particular credit card, and then they can make more money on the other side of the network charging the merchants 12 or 3% fee start by lowering the present one side and raising on the other side even below zero you can make more money than if you thought of them as two separate products. Another counterintuitive thing is often by merging two-sided networks do not only make a lot of money because you play these kinds of interesting games, it actually can be better for society that there is an increase in consumer surplus you’re creating more total value by growing the network. There’s a lot more in the book that works you put in some graphs and charts and lines. You can see them graphically how some of this plays out. That gives them intuition because it is not the answer that you always give stuff away for free. You never give it away for free debit if you understand the economics of two-sided networks you can work through when you do it and when you don’t.>>We had a discussion with our publisher that we have to include a downward sloping man graph in this book. They said there is no way you’re going to include it. We fought that battle and we won it. Another example we one of the Nobel Prize in economics it is playing out in real time right now.>>Check it out. Our editor’s eyes lit up when they saw that one. Another one is playing out right now and is a huge open question is how big a deal is this whole world of distributed ledgers and crypto currencies and tokens in ICO’s >>That is just one thing of this broader phenomenon. It is actually not only economically but intellectually the most interesting thing happening in the business world these days. How big a deal is this distributed alternative going to be. I came out not super interesting going on. Is this going to be a revolutionary thing that will make the country obsolete. I am firmly convinced it is not. There is a theory of the firm that is yielded three Nobel prizes so far?>>yes. A lot of amazingly solid work on this think it’s the notion of what is the company is therefore and how big a threat is this thing called a block chain? It is a super deep issue. We wrote about it in the last bit of the book. I came away thinking this will have some impact and it will make the world of the company as we know it go away.>>I will do two more questions and then there is a mic there. There is an ability to do this online as well. And so by the way I am have plenty more questions. Part of the promise we made Andy and Eric was that we would get a chance to interact with everyone. First one is in the book the one that seemed least developed as an area for recommendations to companies was essentially crowd. This new decentralized or. The whole notion of you can deploy a whole bunch of volunteers, you can use Lenox as a way of getting a whole bunch of volunteer developers. You can have a noncontrolled system which she did a very good description of. How does this work and what is the actual economics of this work. But the map to if I am in it traditional industry other than know this decentralization is coming in the people who figure out are going to have a huge advantage.>>I think that is a particularly compelling quantitative example of how people massively underestimate the power. This will help convince you.>>There are a couple of point things I think any decent size enterprise can do to tap into the power of the crowd. If you have a quantitative tough problem you’re working on where there is an objective benchmark, just not Hayes is a good idea or not, but if you have a tough problem and there is an objective performance benchmark, give it to the crowd. But the crowd work on it. Our buddy is a great scholar about the crowd effort at Harvard. We wrote up a study where the National Institutes of Health said he is our baseline performance for sequencing the genomes of human white blood cells, crowd can you do better? And the crowd came and went give us a little bit on this. They came back 14 days later and they took the runtime from 10,000 seconds to 10 seconds.>>14 days thousandfold improvement. And they took the accuracy up from about 70 to 80% of the total prize money offered to the geek crowd for doing this was $6000. So we said this is the craziest thing you’ve ever seen. He said if you can pose the geeky challenge, you will get a quantum improvement. The other one we got was from a venture capitalist that said for all of history you had a commit a huge amount of now thrown up on Indy Gogo just get a ->>Not just for little companies, in our book we talk about GE not knowing whether this icemaker was going to pay off. Instead of debating it they went ahead to Indy Gogo and they found that they had tremendous demand. The lesson take away isn’t just that you get more eyeballs. That is part of it. But it is more fundamental. It is that the core the people in your organization, they tend to be similar to each other. They have a certain expertise. They are good at doing insurance underwriting or financial analysis or sequencing genomes. There is diminishing returns doing more and more of the same things. You throw it out to the crowd, you don’t just get more eyeballs you get people from totally different fields. You get geologists and petroleum engineers and people who are experts in >>You get people that are marginal to the discipline that pose a problem.>>Many of those other ways don’t help at all. But for some of them it is just a fundamental different way of looking at the problem that is way better. If we relabel some of these variables this is a trivial problem that we saw 15 years ago and problem solved. So you tap into people that just have more diversity of opinion and views and that is something that is much easier to do now and most companies >>You also see how through the core is about the crowd.>>The depression does not love it when you show these results. It is really managerially difficult and subtle to start reorienting the company and not making them feel like they will be fired next week. But they will reach out and do this thing that we hired you to do.>>Now there are platforms of which to do this. One of the things that Cagle is a happy reformulation reshape your question so there is a form that can be addressed. You really need to have something quantitative. But if you have something that you can define precisely enough and people like Anthony Goldblum can help you or say that question is not going to work. But often he will help you reship the question in such a way that you can tap into the power.>>The crowd interfaces a new capability of the companies you might want to spend some time working on.>>We have a long line of questions. I’m going to punt on FAKE NEWS and advertising business models. We might get back to it. But I’m going to wait and do these questions first.>>Hi. I am part of LinkedIn talent solutions. Thank you for joining us. I was at the C sales event in November. I was at the See Sales event in November. One thing that came out for me was pretty provocative. You guys had I felt pretty compelling counterpoints. I think this audience would benefit from your take on that, where he is coming from, where China is going, what we do in the US is a magnitude smaller than what is going on in China. What are the locations for society and possibly the Reformation of government locally?>>Like you just pointed out, our colleague has a lot of exposure to the AI efforts in China. He kinda frames it as an arms race going on and he says China is extremely well-positioned to succeed in this arms race. I don’t know how valuable it is to think about us versus them situation here. He doesn’t feel to me quite like the nuclear arms race. But let’s take that framing and how are we doing here? I had a chance to ask this exact question to Condoleezza Rice and David betray us when they were on a panel. I will channel their answer. They said authoritarian societies really are not great with crazy off-the-wall thinking. It is kind of what they are not set up to foster. Real breakthroughs in this discipline of AI are going to require some fairly radical and innovative thinking. We also have a calling back MIT who cowrote a wonderful book called why nations failed. He says if you have authoritarian states and very economic institutions they are inherently kind of extractive, you should not be long on that? So for those reasons while Kaifu makes excellent points and they will throw a lot of talented people of this problem, I vote on dominance in these geeky fields.>>I will disagree with Andy. I would mostly agree and I was convinced by the general argument. I wanted to be true, but they are becoming more and more worried about things that Kaifu brings up. Just to put some numbers behind what you said, take a guess on the amount of mobile payments on phones in the United States versus China? What if the ratio is? If you took the number of payments through Apple pay versus China, was that what is it? One-to-one? 52 one. The of a thriving mobile payment ecosystem that is just leapfrogging us. I was just at triple AI, the AI conference. Last year the number of submissions of research to AAA I from American versus Chinese authors was about even. This your 50% more submissions to triple AI from Chinese and American research. AAA I used to stand for the American Association they change the name of it. Is not the Association for the advancement. They realize that this is not just an American thing. There are many examples of people pointing out that in a lot of areas AI researchers in China are doing as well or better than researchers in the United States. To agree with bit they were mostly in applied machine learning. Maybe not the most fundamental breakthroughs. There may be some elements that you do I think this is a real challenge but I will end on a point of agreement Andy that I hope you don’t think of this as a nuclear arms race of us versus them. If they figure out a way to cure cancer or Parkinson’s have better self driving cars or whatever, I think that will go down to the benefit of all humanity likewise as we make those breakthroughs if we can maintain an attitude of trade and research and innovation is something that is good for the world and not something >>But that is a bigger challenge. Maintaining those attitudes.>>That is a cultural challenge more than a technological one.>>Morning. Andy and Eric thank you for joining us and thank you for inviting. Two weeks ago there was a company report about the future of labor and labor trends. They were around shifting demographics, baby boomers retiring so it will lower the supply of labor. It was also about automation. And three is around the winding economic inequalities that those two will create. First part is would love to get your thoughts on Crystal ball where you think the workforce is going to be impacted if you ever own production of a percent is going to be displaced, and try to get a sense for that. To, it is music to my ears when he talked about LinkedIn and your challenge on how we can do your challenge on how we can do more. We’ve been thinking a lot about this is a company. We acquired Lynda.com and got into online learning. What can we do to help these low and middle skill workers who will likely be displaced by some of these technologies? What can we do proactively to help them? Would love to get your thoughts.>>Limited the first one. Our colleague Bob Gordon has the best way to phrase what is going on with the labor force these days. He says we do not have a job quantity problem. The robot a public apocalypse that ends brings on massive technological unemployment is just not insight in the data at all. We have a job quality problem. Our colleagues in a beautiful job of driving home the point that our job creation engine used it to get a lot of really good old-fashioned solid American industrial age middle-class jobs. Now the sweet side appears to be less well-paid, fewer benefits, all of that stuff that would Eric and I spend a lot of time on is how we to that engine up and how do we get the sand out of it and give it the best possible chance of getting back into the proper higher gear. I personally think it is way too early to give up and say we need universal basic income because the robots will take all the job. That is an overconfident prediction. It should make us really uncomfortable about making those kinds of predictions about tech unemployment. There are bunch of things that an economist would say we should be doing to try to create the economic environment to let the engines of entrepreneurship and innovation do what they have been doing for 200+ years, which is generate a lot of demand for labor involved skills.>>I agree with that. We talked about that in her book about the kinds of policies. There is a lot we can do to reskill the workforce which is the second part of the question. I did a study with Thomas or who is a professor of the kind of tasks that were most suitable for machine learning. We are very far from artificial general intelligence up there are some very narrow stunningly powerful things that machine learning can do well. MOC has not diffuse the workforce yet that some of it has begun to hit. We have a rubric, a set of test you can do to help. The government is categorize the US economy to 973 occupations and each of those is about 30 different tasks. So if you apply the rubric to the task, it is a time-consuming process, you can get a sense of how the economy is changing. Almost no occupations are entirely affected by this wave of automation. And also almost no occupations are not effective. Our take away is that there is a lot of reorganization reengineering. Talking to the patient, coordinating with the other physicians, the overall plan doing the history all that part is still something that humans do relatively better and even parts of looking at the images there are different types of errors that humans and machines make it is a much richer situation. You can come up with a set of skills that are becoming relatively more important versus less important. This can be used to help shift people in real time or maybe in advance towards those kinds of skills.>>Andy, Eric, thank you for coming. Read thank you so much for moderating so well. Reid thank you so much for moderating so well. For those of you who don’t know we are collaborating with Eric and some of his researchers on a project for the economic graph research program. We arty have results. Stay tuned. There is more great stuff coming. Here is the one question I want to ask. You guys of talked about we have all these amazing technologies are being rolled out here and in lots of other places in Silicon Valley. And yet when we look and nearly all the economic data, they are not suggesting it is suggesting a world that is transforming more slowly than it has for decades, then maybe even a century. Firms are forming more slowly. The rate of occupational changes slowed dramatically. Activity growth is super slow to it is and was a uniform message, just kind of surprising that what you think is driving this disconnect between the very little economic change and very rapid technological change. Do you and us with that gap will narrow, how long do you think it will narrow? I’m curious for your thoughts.>>B wrote an article that we hope will come out called the coming productivity Boone. Would you think it is in the pipeline because the technologies that we see is we walk around we also despite all the investment and say self driving cars, you don’t walk around and see a whole lot on the street, maybe a whole lot on this neighborhood than in the rest of the country. Is not something that has fundamentally changed the productivity of the country yet. People like our friend Bob Gordon say if you extrapolate from the past two years the productivity growth in the future it is bad news. We think that is a huge mistake. You don’t just extrapolate past productivity has zero correlation with future productivity. We showed that in some of our work. The only way to try to make some predictions which is always a pretty risky thing to do is to try to understand one of the fundamental technological drivers. If we look at those we are very impressed. Why are we not seeing more productivity now. We think the answer is this is just the nature of all general-purpose technology. Machine learning artificial intelligence is a technology that affects lots of industries, lots of services. Electricity, the internal combustion engine, in each case it was about 20 to 30 years between when the technology was first developed and started diffusing to when he started seeing a significant productivity booster. We hope you can compress that a lot. I think we are better at doing that. But we have to be realistic that in order to take these amazing technologies and reinvent business models and come up with new products and incorporate them and do all of that we are going to have to take a lot of entrepreneurial innovation. One of the things that does trouble us is that there are a lot more occupational licensing and other sort of sand in the gears of reorganizing the economy right now. It is harder to make some of those changes. So that doesn’t help this process of reorganization but as we’ve been saying, we don’t think the issue is with the technology so much as with the economic and social institutions. One of the reasons we’re focusing much on trying to speed those up is to overcome that abysmal productivity numbers that we are seeing today and speed up the time to address them. It has been a delight working with the team at LinkedIn. We are so fortunate that he US of such amazing data and we are able to figure out how the value of human capital not only affects individual workers wages but in our project we are looking at how it affects the value of firms and a big chunk of the value of the firm depends on the human capital in that firm so therefore it makes sense for firms to invest in their employees human capital and not just say well that is their problem or the government’s problems are and that is probably part of the story that we are putting quantitative data behind.>>We are really confident that productivity growth even with the measurement problems will improve sooner rather than later. I am a lot less confident that business dynamism is going to turn around and start getting healthier in America. People are moving less often, there are fewer businesses new forms. You know who you want to look at at the dynamism has been on the long slow steady decline. The scary part is we don’t quite know why. One of our colleagues has a great description and he talks about death by 1000 cuts which is unpleasant to me because that was I think part of our homework should be trying to figure out how to turn the corner and get our previous levels of dynamism back to the economy. Extrapolating from what you all see around you every day is a deeply bad idea. Silicon Valley is not America.>>I sound like a campaign ad for a candidate that none of us would want to vote for. I apologize for that. Don’t take the wrong message please.>>Thank you for coming out first of all. I am Thomas, a machine learning I deal with this quite a bit. My first part of my question is along the- dust she talks with the cycle of research and how it relates to the industry. The traditional model is you have the US government funding funds to create research. You have companies private, all the Internet companies using research for free essentially and then building and making great profits. First question is how do you think machine learning is being investigated by the US government for benefit of all people and how do you think it is shifting with the evidence of these public offices and more public papers the second question is in terms of education democracy is all based on voters being well informed. How can the common voter be expected to be informed on these decisions and will for policymakers that are also informed. We are having trouble with SEC entitled to regulations right. What are your thoughts on how to better educate people and >>Let me do the latter. I think that is a harder question. I will take the bullet on that one. This is a really tough problem. In both the narrow stance and the broad sense. How do you educate someone to be a decent consumer of fax. Is a really tough problem. I think our ability or our willingness to try to teach critical thinking is heading in exactly the wrong direction. A lot of what we are doing instead is teaching indoctrination which is exactly the wrong thing to do. Finland doesn’t have much of a FAKE NEWS problem. They are right next door to Russia because they’ve done a good job of educating their populists to be able to look at a claim in evaluating say that actually makes no sense at all. It is doable but it is surprisingly hard and I completely agree with you on the urgency of doing it is increasing because the pace of change and the piece of tech is increasing. And it is not the case that the same side of the argument always wins. Sometimes the crazy wins the argument. The official position of the US government is a man-made climate change is not a thing. The crazy one the argument for a long time about vaccines. Crazies winning the argument about genetically modified organizations. When Elon musk talks about summoning the demon with AI I have see what happens when the crazy said Wednesday it is bad news.>>I’m just going to amplify what Andy said. I do think that the answer is very much around educating the population. When we first started writing these books, how do we change things. Here are the things we need to do that basically said I agree with what you’re saying but Alyssa voters want this, I can do it. And I have to go with what the voters want. So for better or worse we live in a democracy and our leaders will to a large extent listen to us. If people are demanding crazy things and they will be pushed in that way. Unfortunately it has to do with getting people to read her books and listen to you and do the kinds of things that we think will change the conversation in that direction. I think it is a way that has to work in society. About the cycle that Prof. Described I think she has a very right that investment in basic RND has been a huge growth engine. Maybe the most important growth engine for a lot of technologies. I think it really is economics 101. There is a public good for basic RND. That is not something that companies will invest in and do on their own. They don’t have the private incentives to do so. So we have to have government investment in the Internet and other basic artificial intelligence. And then fund it in universities and elsewhere. And then I think it is great when companies pick up that thing and develop it and create products and services and they have a private incentive to do that and they become millionaires and billionaires. I think that is more or less the way the system is supposed to work that there is an incentive to pick up that technology and commercialize it. And then ultimately to close the cycle then you have to have a tech system that takes some of that money that has been generated and use it to pay it forward for more RND and research. That is the economic 101 paradigm.>>Thank you.>>Hello, thank you for coming and talking to us. I think we have seen that machine learning is easy to apply it to problems where the objective is clearly standardized. The inputs are standardized. There are a lot of human problems with that is not necessarily true. I think one really good example is the world of online dating right now. There is been a lot of approaches to introducing algorithms, collecting more information from people on both sides. And there has been some success, but a lot of papers show that these algorithms are not significantly better than randomly >>Or just getting drunk in a bar and seeing what happens.>>So the reason I think a lot about this is I am a PM on-the-job steam. The match between a candidate and an employer is actually not that different. I think we like to think that there are a lot of standardized criteria skills that people are looking for but what we found is that recruiters a lot of time use judgment that they cannot explain to talk about a culture fits or someone just clicking. Kind of similar to when online dating sometimes there’s just a spark that can’t be quantified by similar interests. So I would love to hear your take on why you think dating or employing science haven’t been able to crack that much given all of the data we have. Whether that is we have and standardize things properly or we don’t have a complete understanding of what matches. Would love to hear thoughts.>>This literature that you are referring to in great depth, but one thing I do know is that treating finding someone for a job like finding a date looking for that spark and chemistry is demonstrably a terribl way to go about it. Turns out if you are interviewing me for a job, the main thing that makes you likely to hire me as do I remind you of you. Man, that leads to a monoculture and all kinds of discrimination. Get what much more algorithmic approach with a light overlay of is this person a good fit for a culture is right would go. I honestly don’t know if that applies to the spark, the chemistry of finding a life partner. Those are two different activities.>>I have found my heterosexual life partner and I would not use an algorithm to try to get this to happen.>>Let’s leave that there. It was not in a bar by the way. And we were not drunk. I think the answer is somewhere between that I do think the systems can do a lot to improve our decision-making they are not perfect and weak second to just completely fail. This is your person to hire and this is your life partner or whatever. The question is can we do better and can we combine human and machine decision-making and to improve. The example I gave before of the tenure committees is an example of that. There is too little use of data and it is not just that it matches our imperfect I think they are often worse from the human approach than the ones that are supplement by data. It is something that clearly needs more work.>>This is the last question. We are at 24 seconds so we will go little over.>>Hello, thank you for coming in. This is been fascinating. You talked about retraining people make online tools can be retrained the workforce in the speed of change. How do you think about folks that are maybe too old to what extent do you think there will be a population truly has become a set of useless class were folks hit the word but from an economic standpoint folks who don’t have the skills that they can be used in the workforce. How we think about an opportunity for

Leave comment

Your email address will not be published. Required fields are marked with *.