EAC Panel on AI and Jobs
April 7, 2017
March 23, 2017
All Captioned Videos CBMM Special Seminars
CBMM External Advisory Committee Member David Siegel (TwoSigma) talks with MIT principal research scientist Andrew McAfee about the role of artificial intelligence and jobs.
DAVID SIEGEL, CO-CHAIRMAN
Prior to co-founding Two Sigma, David was Chief Technology Officer and Managing Director at Tudor Investment Corporation. After earning his doctorate, David joined D. E. Shaw & Co. and rose to become the company's first chief information officer. While at D. E. Shaw & Co., he founded and served as president of FarSight Financial Services, the world's first integrated personal financial services website, which was acquired by Merrill Lynch. A graduate of Princeton University, David received a PhD in computer science from Massachusetts Institute of Technology, where he studied at its Artificial Intelligence Laboratory. David has had a lifelong interest in building intelligent computational systems and continues to actively pursue this mission at Two Sigma today.
He currently sits on the Board of Directors of the Hamilton Insurance Group, the Scratch Foundation, NYC FIRST and NAF. In addition, David is a trustee of Carnegie Hall, an Executive Advisory Committee Member of the NSF/MIT Center for Brains, Minds & Machines and an advisory board member of Stanford’s Center on Philanthropy and Civil Society.
David is also a member of the Cornell Tech Board of Overseers and a member of the Corporation Visiting Committee for the MIT Media Laboratory.
( Andrew McAfee @amcafee)
A principal research scientist at MIT, studies how digital technologies are changing business, the economy, and society. His 2014 book on these topics,
(co-authored with Erik Brynjolfsson) was a New York Times bestseller and was shortlisted for the Financial Times / McKinsey business book of the year award. He writes The Second Machine Age: Work, Progress, and Prosperity in a Time of Brilliant Technologies academic papers, a blog for the Financial Times, and articles for publications including Harvard Business Review, The Economist, The Wall St. Journal, and The New York Times. He’s talked about his work on The Charlie Rose Show and 60 Minutes, at TED, Davos, the Aspen Ideas Festival, and in front of many other audiences.
He was educated at Harvard and MIT, where he is the co-founder of the Institute’s
Initiative on the Digital Economy.
TOMASO POGGIO: Welcome to all. As you know, this is our third External Advisory Committee meeting. And in our committee we have a set of great friends. I see all of them here with-- they are helping us immensely with their advice on research, organization, and programs.
As you know, the mission of CBMM is research on the science and the engineering of intelligence. And our priority is the science and, as an integral part of it, the education of great young scientists. Hi, Christoph. In this [INAUDIBLE] field studying what I think is the greatest problem in science today-- the problem of intelligence.
So we should-- and we are-- aware of the social implications of studying and, eventually, replicating aspects of human intelligence in machines. In fact, some of us, as Max Tegmark, who may come later, with his Future of Life Institute Foundation, have been pioneers in working to ensure that tomorrow's most powerful technologies, like artificial intelligence, are beneficial for humanity.
So when I saw, a few weeks ago, an op-ed in The New York Times by David Siegel about AI and jobs, essentially, I asked him on behalf of the External Advisory Committee to help us start a discussion within CBMM-- and between CBMM and the rest of MIT, eventually-- about AI and jobs.
And to have an informal discussion, I asked an expert, Andrew McAfee, to help run it. So today's event is, I hope, the preview of a bigger symposium that CBMM may organize a few months from now at MIT.
By the way, as you'll hear today, autonomous driving will threaten a lot of jobs in the US. And so, in the typical dialectic of the luminists we all are, we have a talk tomorrow by another EAC member, Amnon Shashua, on the future of autonomous driving.
So let me introduce you formally, although several of you know him already, both David and Andrew. So David first. David is the co-founder of Two Sigma Investment. I'm quoting to some magazine-- this is a quantitative, trading powerhouse that is the fastest growing hedge fund in the business. Data-driven, absorb large amounts of information to predict the prices of securities ranging from stocks to future contracts.
You are described in this magazine as a computer fanatic, who received a computer science PhD from MIT and worked at hedge funds-- some of the best ones, like D.E. Shaw and Tudor Investment Group.
And the memory I have of David were 20 years ago or more, on the seventh floor of the building over there, where the eye lab was. I remember him with pieces of robots in his hands. He was a friendly and smart presence there, somebody which was great to have around, in the tradition of smart and fun people out in the eye lab.
And then, years afterwards-- I had lost contact with him-- when somebody in the fund business called me to ask me about David. They were doing some extreme vetting of David, of Two Sigma, before investing in them. This was the year 2000, 2001. And I said, he's a good guy. You can trust him.
I was asked at the time because I was involved at the time in a hedge fund-- by far, not as successful as Two Sigma-- but I'm sure that part of the reason I was involved in finance is something I have in common with David. Developing a machine that can beat the market is one of the most stringent Turing tests for intelligence you can think about, or superintelligence. You're competing with other humans.
So, in particular, being able to develop a algorithm that predicts well not only past data but also predict the future of the market is the most difficult task for machine learning. And it's a microcosm of what science is about. Science is about generating predictive models from data.
David-- so we share that. And Andrew-- Andrew is a great guy who has done great work across the street as well in school. And I always look forward to conferences where he is speaking because it's always, not only fun, but is extremely interesting. Has been studying digital economy, how technologies are changing business and society. He's the co-author with Eric Sohn-- of course, I know Eric. But pronouncing his name is still difficult-- of the 2014 best seller of The Second Machine Age and the more recent Race Against the Machine.
He received his PhD from Harvard Business School, or a doctorate and two Masters of Science and two Bachelor of Science at MIT. So his blog is highly recommended. He writes for a number of publications, including Financial Times, The Economist, and so on. So I'm sure I can put all of this in your hands and it will be very interesting. Thank you for coming.
ANDREW MCAFEE: [INAUDIBLE]
Tommy, thank you for that. Tommy asked me to kick off very quickly. But we're going to segue, as soon as I can manage it, into a conversation between me and David, first of all, because I've had questions I've wanted to ask him for a long time. But then we're going to open it up and we're going to talk as a group about some of these issues.
I want to start off by echoing just a couple of things that Tommy said. First of all, Tommy, I agree that unlocking consciousness is the greatest problem in science these days. What's interesting to me, because I'm not a scientist of intelligence-- or of intelligence, according to a lot of my colleagues-- what's fascinating to me is that as we unlock this challenge progressively, the economic and the social consequences of getting better at figuring out intelligence and building machines that can do more and more intelligence-like things-- it appears to a lot of us that those consequences are profound. And I'm absolutely of that view.
In academia, we talk about doing cross-disciplinary work a lot. And I put air quotes around that to indicate that it doesn't happen very often. The CBMM is one place where it absolutely happens. What I've learned, though, over the past couple of years is this topic-- about what's happening-- are the robot-- I'll say it in a flip way-- are the robots actually eating all of the jobs-- this topic actually draws in a truly cross-disciplinary group from throughout the Institute.
And there is-- I say this with some hometown pride-- there is no better place on the planet to have these conversations than MIT. Because whatever I find myself wanting to get smarter about, one of the world's best and most experienced people on the topic is down the hallway or across campus. It's an absolute playground for these issues.
I'm going to kick off today by doing something that I feel is in the best MIT fashion. I'm just going to bombard you all with a bunch of data. Because that's kind of how we like to learn here. When I'm done with that, then we're going to take it easy and we're going to segue into a conversation with David.
And what I tried to do-- and this is an echo of a presentation I gave at a conference that Tommy and I and a couple other people were at a while back-- was to try to assemble an initial frequently asked questions on what's going on here at this intersection of tech progress and jobs and wages.
So the first question that I get asked all the time is, OK, what, actually, is going on at this intersection? And the only honest answer is a whole lot of things are. To me, one of the central phenomenon is what's been called the polarization or the hollowing out of the workforce in America and lots of other countries.
So to show some data on that, I want to rely on work that was done by our MIT colleague David Autor in the economics department, who's just done a large volume of wonderful work on questions having to do with the labor force-- around the world, but particularly in America.
This is a paper that he published in 2015. And I'm going to recreate my favorite data representation, my favorite graph, that David drew in this. Because I think he took a brilliant approach to helping us understand how the American workforce is changing.
And what he did was essentially look at all the jobs in the United States in 1979. And he line them up from lowest-paid to highest-paid. And then, right there in the middle, is-- duh-- the American middle class. And in 1979, we had a really large, stable, prosperous middle class. I look at it as one of the crowning achievements of American society and the society of other rich countries around the world.
And then, what David did was, he then played the tape forward in more recent years. And what he was looking at was not how the raw volume of jobs evolved but, if you like, how the market share of jobs was changing from time period to time period. And what that led to was lines that look like this.
So this is what happened between 1979 and 1989. And the way to think about this is every time you see an area below that red line, that's where the market share of those kinds of jobs went down. So there are proportionally fewer Americans doing those kinds of jobs. Every time you see an area above that red line, that's where the market share of jobs increased. So does that intuition make sense? God, I love MIT. You can just do that very easily.
ANDREW MCAFEE: No. That's some flavor of typo. I'm not deliberately elongating that axis. I have no idea what happened there. And you have very sharp eyes.
AUDIENCE: I think-- [INAUDIBLE]
ANDREW MCAFEE: Ah. It's a half zero. You have even sharper eyes. OK. So ignore the final number on the axis there. So look at that picture-- between '79 and '89, I find this a really happy story. Because the American economy was proportionally losing jobs down there at the lower levels of pay and doing a really good job of creating middle class, upper middle class, pretty high end jobs. I find this a pretty happy story.
And then what I want to do is draw more versions of this line for more recent time periods. And then you're going to start to notice something different. So keep your eyes primarily on that bulge of where the jobs are being-- where market share is going away-- what kinds of jobs are we proportionally losing. And you're going to notice one fairly consistent phenomenon.
So now let's go from '89 to '99. And you start to see that bulge of job loss is moving upward in the wage range, in the skill range, starting to eat kind of into the heart of the middle class. And we started to create these lousy jobs down there at the absolute lowest pay range. You know, there's still some pretty good jobs being created here.
The next time period I find very strange. This is what happened between '99 and 2000. And it's a story of complete blech in the economy, except for generating a lot of very lousy jobs down at the fairly low end. So the rest of the economy kind of flatlined. And we were generating lots of jobs down there.
In other work, David has looked into what kind of jobs those were. These are not low-end manufacturing jobs. These are in-person service jobs. I always think of home health aide as the classic job somewhere down there. Very important to do right. Despite the work of people in this room, we don't have robots that can do that yet. But it's an $8 to $10 an hour, no set schedule, no benefits, no health care kind of job. So I put it down there in that fairly lousy category.
Yep. So this is stable at the 1979 distribution of jobs slash wage-- wage or skill, which are interchangeable here. No other audience asks questions at this level of detail. This makes me incredibly happy. And then, for the most recent years we have data, this starts to make a little bit more sense to me. We're starting to see the high-end jobs come back.
So this is the data scientist, user experience designer economy. There's a lot of really good, high-end, upper middle class, upper class jobs being created. But look at that bulge of where the job loss is happening. For me, this is right squarely in the heart of the classic American middle class. So I love this representation because it just shows what's been happening in the workforce over time.
After I try to communicate this, the most common question that I get is, OK, are we deeply weird as an economy? And it turns out that we're not. Because, thanks to some other interesting work-- no-- thanks to some other interesting work, if you look across a large group of European companies, this is a different way to look at the same kind of data.
So for every one of these countries, there were three lines-- low-pay jobs, middle-pay jobs, high-pay jobs. Let's look at how that shifted around, in this case between '93 and 2010. And the phenomenon is incredibly consistent. All of these countries lost jobs in the middle class. There is a mix of whether they gained them down low or up high. But this hollowing out of the middle class is an extremely consistent phenomenon.
The next question I get asked all the time is, OK, is this a China an outsourcing, a globalization story? Yes, to some extent. But I actually think the main answer is, no, it's not.
Let me show you one way that I think about this. This is a graph of America's manufacturing output over time. And what you learn from this is that the crisis in American manufacturing has nothing to do with output. We are a large and growing manufacturing economy over a pretty long period of time.
China is now the world's largest manufacturer. We're number two and nobody else is coming close. Our manufacturing output these days is as large as that of Germany, India, France, and Italy combined. We're a big manufacturing economy.
ANDREW MCAFEE: A-ha! That's what we call a segue. Because the next line I'm going to add is US manufacturing jobs. And the year of peak manufacturing employment in the United States was 1979. So the kind of easy conversation about the crisis in manufacturing is, if anything, it's a jobs crisis not an output crisis.
And you just see this-- in recent years, the most recent data shows that this little happy rise is actually tailing off and heading in the other direction as well. So we do not have an output crisis. We have an employment situation that's changing very quickly.
David and a colleague, again, looked at this situation and said, OK, what is going on? And I place a lot of weight on what David says for a couple reasons. One-- he's a first-rate economist. Second of all, he is not as big-- I see a computer under every rock. I see a robot everywhere I look. David doesn't.
So when he comes to a conclusion like this-- when he looked at what was happening with the shift in jobs in America-- when he comes to a conclusion like this, it makes me really happy because he makes the point-- correctly-- that the American middle class was built on the back of routine work-- routine physical work, assembly line, routine cognitive work, payroll clerk.
We have robots and automation and pieces of software that are really good at all that routine work. And that explains the changes that we've been seeing in the labor force. David and his colleague looked at a bunch of other factors. They did really lovely work here. And they said, look, these other things they don't explain-- they don't appear to be central to the story that's going on here.
This is a story about fairly-- you know, by the standards of folks in this room-- fairly dumb pieces of technology, substituting for routine work throughout a pretty big, industrialized economy.
The next question I get asked all the time is, OK, you're telling me a story about robots eating all of the jobs and yet we have less than 5% unemployment in America. What's going on? Something doesn't seem to make sense. And the robots eating jobs is a pretty naive way to look at it. What any economist would say is, if the price of something goes down-- the price of hiring somebody-- or if demand for that thing goes down-- it's not the case that we use absolutely none of it. It's just the case you're willing to pay less for it.
And, in particular, if we have a combination of factors, you're not going to have mass unemployment. You're going to have something else. So if your labor force is fairly flexible, and your economy is growing, that's kind of a recipe for job growth. And we should keep in mind, the American economy has been adding net jobs month by month every month for, I believe, more than six years now. So we're not in any mass unemployment crisis.
We have the most flexible labor force across the rich countries club of the [AUDIO OUT]. And what that means is that-- the top line there is just raw economic growth, just real GDP per capita.
And what you notice is, as the economy grows, the raw number of jobs grows almost in lockstep with that. It's almost a one-to-one relationship. The gap between those two lines is just greater productivity over a time. So we're absolutely adding jobs.
The problem, when you hollow out the middle class, is that, on average, the jobs that you're creating are not as good as they used to be. So if you look at that 50th percentile American family or household, you see that it has really stalled out.
There's a little bit of encouraging news in recent years. But the American 50th [AUDIO OUT] family has been stuck, since at least the turn of the century, at a flat level of income. Probably, as we look harder, the situation looks even worse than that.
There's one little wrinkle in this story about why we don't have mass unemployment. There actually is one concentrated demographic where we do have pretty serious unemployment in America. And, again, it is right there in what I think of as the heart of the American middle class-- it is prime age men who grew up during kind of brawny, you know, stereotypically masculine labor in factories-- those jobs go away.
What those people do not do, in general, is take the job as a home health aide. What they do instead is start sitting on the sidelines of the economy. So this measures the percentage of prime age American men who either have a job or are looking for a job. That's labor force participation, if you're in one of those two categories.
If you're not, you're part of this falling off here. Fully 12% of prime age US [AUDIO OUT] don't have a job and are not looking for a job. They're purely on the sidelines of the economy.
Like we can imagine, that is even more concentrated among the least educated, prime age American men. So very close to 20% of high school or less educated American men are absolutely on the sidelines of the economy. And that I consider to be, actually, mass unemployment among a pretty important demographic. There are a lot of these people, for one thing.
Last thing I want to talk about is-- second-to-last thing--
ANDREW MCAFEE: It depends on age. It depends really heavily on age. So if you're talking about men in their 40s and 50s, most of them-- this is a little bit of speculation-- I would imagine most of them had a job earlier in their careers. One of the weirder phenomenon is American men, especially less educated, starting their-- what I used to think of as their careers-- but never having [AUDIO OUT] all.
There has been some research that came out of the University of Chicago that I find pretty crazy. He looked pretty hard at American men in their 20s. He found a mass unemployment, or labor force participation crisis. And he looked at what they actually did with their time.
AUDIENCE: Video games.
ANDREW MCAFEE: Yep. Two words-- video games. They [AUDIO OUT] games an absurd number of hours a week. And when you ask them about their life satisfaction, they're thrilled. They're having a great time. I'm going to show you some data in a minute that points in the other direction.
But when you ask these people-- they say, it's great. I love my World of Warcraft people. Anyway, yeah, please--
ANDREW MCAFEE: Wow. This is the best segue audience I've ever had.
So let me talk about the consequences and the causes of it, because they're all bundled together. And I want to be really careful, based [AUDIO OUT] just said. It is tempting and incorrect to tell very simple stories about cause and effect here. There are a bunch of rich phenomenon. They're all intersecting and feeding off each other. And economists love to talk about causal density, which is their fancy way of saying there's a lot going on here.
So I want to emphasize, there's a lot going on here. And I just want to talk about correlations instead of clear cause and effects. We don't-- it would be very hard to assess clear cause and effect. And I don't think the research has done that yet. So I want to show some correlations and then we can-- David and I and all of us-- can speculate about what we think the causes are.
But let me share some other things that are going on at the same time as these phenomena. Please.
ANDREW MCAFEE: So this is the number that I walk around with about people who are out of the labor force-- just not even trying to participate anymore. So what are some of the consequences here? And let me say this one more time-- I'm going to show some things that are correlated with lots of technology and computers proliferating. I am not saying that computers are causing all of these things.
I find it important to be clear about this because I'm going to show some data that, quite frankly, I find terrifying. One of the things that's going on is a recently understood rise in mortality that was unexpected, is bizarrely large, and is extremely bad news.
Angus Deaton won the [AUDIO OUT] Prize in 2015. And he and Susan Case-- instead of going on vacation for four years, which is what I think I'd do-- they dove into the data on mortality among different groups in America. And one of the happiest phenomena on the planet is that almost no matter what demographic group you look at in any part of the world, [AUDIO OUT] rates-- mortality are going down. They're going down very quickly. The world is getting a lot healthier.
Deaton and Case found an exception to that, which just makes my jaw drop. This is research they just published. I'm going to reproduce a graph that came out in the Journal today or yesterday. This is mortality-- this is only America-- this is mortality for African-Americans-- and, again, going down-- we like that-- Hispanic-- going down-- we like that.
These are least educated whites in America. Their death rates are going up-- so much so that for-- this line is causing the mortality rate [AUDIO OUT] white Americans to go up. This is a reversal of the trend that was happening before. We were not aware of it. We were not expecting it. And it's happening no matter what age we're looking at.
So I don't know how happy those people are, but they're clearly dying at a higher rate. This is extremely atypical, compared to other wealthy countries-- even [AUDIO OUT] countries are getting healthier quite quickly. US whites have a completely different pattern than populations in other wealthy countries.
And the causes are really tightly clustered. These are not due primarily to obesity. They're not due to poor diet. They're due to three main causes-- suicide, alcohol and drug-- acute alcohol and drug intoxication-- overdosing, and chronic liver disease.
AUDIENCE: [INAUDIBLE] --that segment-- [INAUDIBLE]
ANDREW MCAFEE: It's the-- white Americans are 62% of the American population. So let me say that a different way-- If you look at the difference in [AUDIO OUT] mortality trend for the least educated white Americans versus what's actually been happening, the change between the old trend and what's actually happened is larger in magnitude than the AIDS epidemic.
So, for example, there were more opioid deaths due to overdose in America last year than there were deaths in America from AIDS at the peak of the epidemic in 1995. So these numbers add up very quickly. And it's an absolute public health crisis. If that's not distressing enough to everybody, I've got more good news for you.
There's other really interesting work about what Charles Murray has called the coming apart of American [AUDIO OUT] And we can disagree with some of his views-- he's an extremely careful digger into different social data. So let me show some graphs that he put in his book Coming Apart.
He had this really clever idea to not think of the American middle class as one big, homogeneous group, but to look at what's been happening, essentially, at the top 20% of the American middle class versus the bottom 30%. So, you know, upper middle class families like a lot of ours, frankly, and then lower middle class families.
He stereotypes them. He calls the upper middle class town-- the archetypal town-- Belmont and [AUDIO OUT] middle class whites live. And he looks at things like the divorce rates.
And you're going to see this very common pattern across the data here-- upper middle class white Americans-- the residents of Belmont-- live just like they did a half a century ago. We are still in kind of Eisenhower, Leave it to Beaver- era America. Lower middle class American families are living substantially different lives, and usually in directions that terrify me.
So divorce rates are going way up. In Fishtown, about a third of kids now are living with both of their biological parents. In Belmont, that rate has been steady for 30 or 40 years. The white imprisonment rate has skyrocketed among lower middle class Americans, while none of us are going to jail anytime soon.
Just for context, the reason Murray was concentrating on white Americans here-- I think he was tired of getting called a racist. So he was trying to show that the phenomena he were interested in were not just about African-Americans.
One other thing that's going on here is a rise in what this book calls "the politics of resentment." This I found a fascinating book. This is a sociologist who, for a few years now, started hanging out-- she's from Wisconsin, she's at Madison-- she started hanging out in in rural, white, middle, [AUDIO OUT] middle class communities in Wisconsin.
And early in 2016, before the presidential campaign heated up at all, she published this book and talked about just the simmering, ongoing resentment that she kept coming across among people in these communities, who were not economically devastated, but they just felt like things had been changed, the bargain had been perverted for them, and that she kept on feeling this overwhelming sense of resentment.
Where does that take us? That's Nigel Farage, who was the head of the Brexit vote in the UK and that's our president right now. I don't have to look too hard before I find signs that societies where the middle class has been getting hollowed out tend to do more populist things when they have a chance to express their resentment via the vote. All right. That's the end of the politics here.
The last question that I want to talk about are, are things going to get better? And there's a really stark difference in opinion among some of the smartest people that I get to talk to [AUDIO OUT] topic. And a lot of very smart people say, look, yes. You're falling into the trap of being a Luddite 200 years after the fact. And what we know about the forces of entrepreneurship and innovation and capitalism is that they find new things for people to do.
And the pattern for [AUDIO OUT] years of the industrial era has been a very clear pattern-- steadily-- very steadily-- rising wages, standards of living, incomes for average workers over time. Their kids did better than they did. And, yeah, we're going through a rough period now. But if we make some tweaks to the system, maybe, and make the engines of innovation and entrepreneurship work better, we can get back to a better place.
The argument from history would say that things are going to be fine. The argument on the other side says absolutely not. So you can either underestimate the power of capitalism or you can underestimate the power of technology. I think they're both very, very strong forces.
The argument on the no side is that we're underestimating how powerful technology is. And, in particular, what we are now able to do-- which is build technologies that escape that box of routine work, where they've been for kind of the entire computer era, and start going out and doing things that we used to need human beings to do, including very highly skilled, very intelligent human beings-- computers are better at that stuff now.
So this view kind of says that you ain't, we ain't, seen nothing yet, with technology's impact on the labor force, on the economy, and, probably, on society at large. With that, I want to stop talking. David, would you mind joining me and let's carry this conversation on. Hey, Mark. Yeah, sure.
ANDREW MCAFEE: Yeah, thank you for that question. This is a talk geared for a very specific audience-- the one sitting in this room. I don't feel like I need to tell or try to communicate to this group that tech progress is a beneficial force.
So I was concentrating on some of the challenging consequences of it. I couldn't agree more with you that robots and computers and AI and these other things are essential for us to do. They're incredibly powerful [AUDIO OUT]. They will make us wealthier overall.
The challenge, I believe, is that the way we used to divvy up that wealth is shifting kind of quickly.
AUDIENCE: [INAUDIBLE] to be pure and simple enough-- [INAUDIBLE]
ANDREW MCAFEE: Every economist that I talk to says that technological progress is the only free lunch that they believe in. And what they mean by that is there's no other route to higher productivity than going through technological progress. And higher productivity, by definition, is awesome because it lets us do more with less.
ANDREW MCAFEE: And why might higher productivity be bad? And one reason-- and, David, come on, let's sit down and talk.
DAVID SIEGEL: You know, what I would say to that question is-- and maybe we should turn off the projector-- because--
ANDREW MCAFEE: As fond as I am of our book cover, I don't think we need it in the background all the time.
DAVID SIEGEL: I think, really, that the discussion has more to do with [AUDIO OUT] work. And, really even more basically, the future of what people will be doing with their time. And so all this progress is all good. But, on the other hand-- and, you know, there's much to say here-- other people have to also consider and think about the adjustment that will be needed in the world to essentially satisfy the needs of human beings.
And to do that-- and one of the reasons-- you asked me earlier, why did I begin to start to think about this problem, which is very much related to your question-- so I'm in an AI background, I'm in the investment world.
In a way, I kind of got pushed into this problem by people peppering me with questions. They expect [AUDIO OUT] something about this. Because I was thinking about investing, thinking about the power of using AI to make investments, and that, somehow, I would have predictions about all of this going forward because of how I'm sort of in this particular circle.
And, you know, quite frankly, I didn't have any views whatsoever. But I started to wonder and-- for example, this is something that really, I think, amplified the problem we have. So one of the things we do is venture investing in the tech sector. And so I'm looking at venture deal after [AUDIO OUT] and, you know, 90% of them are about automating away some kind of human activity.
And if you then start to read about how companies deploy capital-- so the investment world is accused, [AUDIO OUT] so, of being very biased to short-term quarterly improvements in performance of a business. And, generally, the best way to get a better quarterly result is through tried and true automation-- to invent something new, to create a new market, that's actually pretty hard. And you're not going to do that in, what, in a couple of months.
But, on the other hand, you could deploy some new software or you could put a robot on your assembly line and you're almost guaranteed to get--
ANDREW MCAFEE: Costs go down, margin goes up. Things look better.
DAVID SIEGEL: That's right. And so as I start to digest sort of the economic world that I'm in, I'm concluding that we have an imbalance, where, basically, in some very strange way, it appears that we're all collectively trying to eliminate human work. And that no one, in particular, is thinking, well, what would be things to replace the activity with.
And there appear to be no-- in fact, the economic incentives are, again, to replace human work.
ANDREW MCAFEE: Let me ask you-- I have a mic on-- let me ask you the toughest question that I get asked all the time, because I'm going to steal your answer to it. What about the fact that for 200 years a succession of allegedly smart people has been saying some version of what you just said and what I believe as well, and for 200 years, we have been dead, flat wrong. And what we keep on underestimating is the fact that all these new technologies, all of these new companies, all these new [AUDIO OUT] in aggregate, need a lot of people? And provide enough new jobs and good livings?
DAVID SIEGEL: Well, you know, first of all, I think it goes back to the nature of work. So let me just make a simple analogy about [AUDIO OUT] Imagine that you have an economy of dogs, right? And so dogs-- I don't have one but I used to-- have certain unmet needs. They need dog food. They need a flea collar. You know, they need certain things.
But, after a while, you know, the dog's life is good. They don't need anything more. I mean, you can try to give a dog an iPhone, but they're not going to appreciate it. And I think it would be very difficult, if our jobs were to create a growing economy around, you know, ever-increasing dog GDP. Because dogs are dogs, right? I mean, it makes sense. Well, it--
DAVID SIEGEL: Well, we could. That's a good idea. We can do that. Well, actually, in New York City, in fact, we specialize in increasing the GDP around dogs. But, you know, it's not really going to go very far.
ANDREW MCAFEE: That might not spread too far outside the Upper East Side?
DAVID SIEGEL: You know, we have like the dog walkers. We can automate that-- maybe robot dog walkers. But, the point being that-- so you've made the horse example, where eventually horses are no longer-- you're probably familiar with that-- you know, I'm going at it from the other way around.
ANDREW MCAFEE: From the demand side.
DAVID SIEGEL: From the demand side. And so, I think-- so that's a dog. Let's talk about humans. Humans are like a dog, in that we have, actually, a limit, I believe, which people are not being very candid about-- a limit to our own demand. So, for example, there are only so many electronic gadgets that you want to own. There are only so many books you can read. There are only so many vacations you can go on, so many back rubs you can get.
So, eventually, you hit a point where we are overwhelmed with only so much food [AUDIO OUT] to eat. And so on the demand side, you know, we've done a great job in our economy. And we've actually perfected the automation to make the demand efficient.
And so the question is-- and I think this is-- I strongly believe-- that we're actually [AUDIO OUT] the US, for most of the population, we are basically meeting the needs and the demand of humans.
And so unless we evolve into some radically different species, where we have eyes on the back of our heads and in the front and we can watch two movies at the same time, there is a limit to this. And so I think, in part, this is why so much of the venture activity that I see is focused on substitution. Because, in fact, you know, they're not going to really invent something new that we're going to do, because no one has the time to do it.
ANDREW MCAFEE: It's just easier for them to say, I can [AUDIO OUT] --ologist away.
DAVID SIEGEL: That's right.
ANDREW MCAFEE: OK.
DAVID SIEGEL: And much harder to invent a new activity. Because, first of all, that new activity would have to-- because it has to steal your time from some other activity-- it's a bigger quantum to get people to adopt it.
ANDREW MCAFEE: Let's have the rule of not having any rules. I would love to keep this conversation as broad as possible. I have questions I want to pepper you with. But let's-- please--
DAVID SIEGEL: You know, the universal basic income concept is favored on the west coast. And another thing that, by the way, that drew me into this debate was when I meet with my fellow tech crowd, the entrepreneurial crowd, I'm very surprised at how unconcerned they are with this problem.
ANDREW MCAFEE: They're a little blithe about the whole thing, aren't they?
DAVID SIEGEL: They really are. And--
ANDREW MCAFEE: Can I paraphrase?
DAVID SIEGEL: Yeah, please.
ANDREW MCAFEE: They kind of think that mass joblessness is coming. But it will be all burning [AUDIO OUT] all the time. It's just going to kind of feel-- we're just going to go party all the-- is that about right?
DAVID SIEGEL: That's right. Or, maybe, we'll just give people endless entertainment and somehow that will let them eat cake kind of thing. But, what I notice is, the people who are suggesting this are the people with rewarding jobs. And so--
ANDREW MCAFEE: They've got some of the coolest jobs in the world, right?
DAVID SIEGEL: And so I don't think that they're-- and, look, we have evolved over how many years? I mean--
ANDREW MCAFEE: Lots.
DAVID SIEGEL: --millions, really. And, [AUDIO OUT] I think we are-- and this is beyond the scope of my knowledge-- but I suspect we are hard-wired to want to work. Because we wouldn't have actually survived if that were not the case. So we probably are going against our basic genetic programming, if we are basically going to tell people, now you can just sit around all day. And you don't have to do anything and it'll be just fine and dandy.
ANDREW MCAFEE: We kind of rebel against that at a genetic level.
DAVID SIEGEL: I don't-- do you want to do that? I don't want to do that.
ANDREW MCAFEE: Not for more than a couple days.
DAVID SIEGEL: I'm not so sure it will be a big problem taking care of the people, at least, depending--
DAVID SIEGEL: I agree with you that that-- but the question I would ask-- OK, that's an implementation problem, and it's very worthy of debate. And I don't think we're doing the right things. And I could make some suggestions, but it's a complicated problem.
But I would just start with the question of, are we able to do that? Is there, for example, enough wealth in America-- if we distributed it the right way, and if people weren't squandering it on the wrong activities, and so on-- could we basically have everyone-- by the way, you need a measure of success. So what is the metric that we're using here? Is it GDP?
I would argue that, by the way, that's part of what's causing the problem-- that GDP is not the right measure to be focused on. Is it a happiness index? So part of this is a measurement issue. And I think that we do not have in our society the right metrics for gauging-- the charts that you put up, I think, are interesting and clearly bad indicators.
But, overall, how would you come up with a couple of numbers to indicate whether or not an economic policy is really what society wants?
ANDREW MCAFEE: And it's interesting-- our economic dashboard is pretty bad because that mortality increase was unknown until a year or two ago. So the dashboard that we were looking at didn't pick up on that pretty important signal.
DAVID SIEGEL: That's interesting. But then you could also-- but, again, these are, again, important questions for society, given the era that we're in-- arguably, we could say, the most important thing to do in society is to extend life and that we should be devoting more and more resources to that problem. Because now that no one's starving-- or, maybe, you know, well, we should be expending more money, actually, on virtual reality. Because even though your life may not be so long, you'll enjoy it more.
And so this question of how to-- to your point-- how do we redefine not only the future of work, I would say, but related to this is how society allocates its resources to solve what it perceives are the important problems.
And that is changing, given the fact that most of our basic unmet needs are met.
ANDREW MCAFEE: Are now getting met. Hold on a second. Go ahead, please.
Let me-- this is a fundamental [AUDIO OUT] on that. Why-- I get asked this question all the time, so, again, I'm putting this on you-- where does your intuition or your confidence come from that we're heading into an era that's characterized more by substitution of automation for people than by complementarity, than by like awesome robot surgery but with a person still there?
DAVID SIEGEL: Well, I think, in the business world, I think that, certainly from what I've observed, people have found that it is easier to just substitute, to think of the problem as a substitution problem than as a [AUDIO OUT]
Obviously, it sounds much nicer to say that you're going to be building software that will just empower the human workers to do better.
ANDREW MCAFEE: Which is what every CEO I've ever talked to says.
DAVID SIEGEL: They're not telling you the truth. Because, in fact, that problem actually-- I mean, that's hand-waving. So, in other words, what they're saying is they're going to design a whole new set of tools that sort of an average skilled worker can now use these incredible tools and achieve much, much more.
So, on the other hand, what if you could design the tool to just completely replace the worker and be done with it. So you're going to first try to do that. And, again, this goes back to the nature of work. So the tasks that people-- to your point-- the tasks that people are doing in business really aren't actually changing that much. Yeah?
AUDIENCE: [INAUDIBLE] robots.
ANDREW MCAFEE: Mark, how far away is that robot?
AUDIENCE: People are working on it [INAUDIBLE].
ANDREW MCAFEE: Can you ballpark that for us? Is it a five-year or a ten-year problem? Five-ish. Wow.
DAVID SIEGEL: But, again, to your earlier point-- what stops a lot of the end mile automation is simply that the human labor becomes very cheap. And so the last thing [AUDIO OUT] automated--
ANDREW MCAFEE: Are the cheap stuff.
DAVID SIEGEL: --are the cheap stuff that humans can do. And, really, I mean, humans are actually, at some level, very flexible for an unstructured environment. And robots will get there. You know, Mark is speeding up the advances here. It will happen sooner than people think.
But, clearly, [AUDIO OUT] I think, humans are increasingly going to be thrown into the jobs that Mark Raber just--
ANDREW MCAFEE: That's left over.
DAVID SIEGEL: --that he didn't get to yet.
ANDREW MCAFEE: Is my intuition-- can I say one-- can I ask one quick follow-up question? Is my intuition correct that you and your colleagues are making faster progress than you used to at having these machines do useful things in the real world?
But I'm curious. You said like the robot on the truck-- loading-unloading robots, five-ish years away. That sounds like fast progress.
DAVID SIEGEL: But, Mark, even though it's true that compared to a human, you know, mechanism, robots are still primitive. But, you know, the nature of-- again, we're not talking here about a theoretical problem of, in a specially designed competition, who's more flexible-- a human or a robot? OK. It will be a long time [AUDIO OUT] robots get there.
What we're actually talking about is the nature of work, OK? And what is actually going on today. And so, as I think you know, for most applications, the activities are surprisingly easy to automate. And even in China right now-- things that were [AUDIO OUT] not automated because labor was so cheap are very rapidly being automated.
In Wall Street, hundreds of thousands of jobs in middle office and back office functions-- again, to your point-- that were middle class jobs-- people adding up numbers, people updating ledgers, and so on-- these jobs have silently gone away over the past 20 years.
And what have they been replaced with? They've been replaced with home health care workers, and so on. Because the automation needed for Wall Street, it turns out, is actually pretty simple stuff. And in most cases, by the way, we're not talking about AI. We're just talking about some decent software.
So I've spent a little bit of time exploring exactly [AUDIO OUT] question with officials from the government, officials from industry, and so on. And so I've come up with a very depressing conclusion-- so, number one-- the people, like the people in this room, that actually have the ability to [AUDIO OUT] very clearly about trends and-- by the way, a lot of the problem is, you have people like Kurzweil, who are exaggerating things to a point where it creates mass confusion. And then policymakers and others don't even know what [AUDIO OUT]
So I think that he's-- I know him. He's a smart guy. But I don't think he's being helpful in making policy decisions. So then you have the people, the policy makers, the people that are running major businesses, and even industrial businesses, that, actually, sort of lack the perspective that every one of you have in this room.
And the gap is especially large in the government. And so when, for example, the government is trying to formulate trade policies, tax policy-- they should read some of David Autur's work.
Another thing-- I'm jumping around a bit, but very much related-- it is surprising to me-- and, you know, this was part of my own path to my thinking on the matter-- how little basic research people are doing on the problem. David is one of the few--
ANDREW MCAFEE: And he's been a great exception for a long time.
DAVID SIEGEL: That's right.
ANDREW MCAFEE: We were kind of slow to this party. And so I think that, if I were to make a few suggestions-- first of all, more basic research is needed, economic research is needed, in these trends.
Secondly, the community of people that are involved with the development of these technologies have to offer more level-headed assessments as to where the technologies are going-- to not confuse the people that are making policy decisions. And, you know, the list is much longer but I'll start with those two.
ANDREW MCAFEE: Bob, and then Tommy, please.
DAVID SIEGEL: You know, some of it is a [AUDIO OUT] of a culture. And so, if a culture-- I don't know anything at all. I'm really just making this up. So full disclaimer here. But different societies-- and, you know, I've traveled the world-- when you look at Japan versus America versus Germany versus-- the nature of work differs considerably in each of these societies.
And so I would not immediately conclude that we can not shape, perhaps with-- in fact, not perhaps with-- with government [AUDIO OUT] and this is, I think, a key thing-- that if we want to change the nature of work and the kinds of jobs that we're creating, there are a variety of levers that the government has-- through taxation policy, rules and regulations-- there's plenty of ways the government can push society in certain directions.
And so, ultimately, if we all agree that robots will do most of our manufacturing-- and I think we have to agree that that's the case-- then you're going to rapidly come to the conclusion that humans have to be more involved with non-manufacturing activity and non-routine accounting operations.
Well, what does that leave you with? It leaves you with the kind of service economy that actually, luckily, the United States has a pretty strong one. But what this then means is that we have to encourage the creation of service jobs. We have to get rid of the stigma--
ANDREW MCAFEE: And make them stink less basically.
DAVID SIEGEL: Make them stink-- exactly.
ANDREW MCAFEE: And give them some stability, give them a packa-- make them look like classic, industrial-era jobs that have that nice package of stuff.
DAVID SIEGEL: And, by the way, in Japan, I would argue-- and I'm not saying that they have a great economy-- but service jobs in Japan are more valued than in the US. But I would say here's, really, the bad news in America today-- I think-- and I've seen surveys on this that back it up-- that, in general, people-- consumers-- would prefer to interact with automation over the current level of human worker that they encounter.
And so you see, this is, again, part of the diabolical race to the bottom.
The data says otherwise. So the data says that when you basically-- when people don't have jobs, they--
ANDREW MCAFEE: They kill themselves.
DAVID SIEGEL: They kill themselves.
ANDREW MCAFEE: To be slightly oversimplif--
DAVID SIEGEL: That's right. Or kill other people.
ANDREW MCAFEE: I think we do need to be careful about very simple cause-and-effect relationships here. There is something about lives and families in communities where the structures of the industrial era have gone away that lead to some worrying things. And to Tommy's--
DAVID SIEGEL: But, in particular, here is, actually, my worry about this pathway. Up to now, and for thousands and thousands of years, we have distributed the wealth of society based primarily-- but not always-- on work that people do.
ANDREW MCAFEE: On labor income.
DAVID SIEGEL: Labor income. And that's how we decide who gets what. And now you can not just wave your hand and say, well, now we're going to switch over the next 10 years to a system where some government bureaucrat--
ANDREW MCAFEE: Where something else happens.
DAVID SIEGEL: --is going to decide how much money you get in your UBI. Because who will make that decision? And so I think the thinking here is very naive because, essentially, what you're trying to do is to restructure something that we've perfected over 10,000 years in 20 years.
ANDREW MCAFEE: Yeah.
DAVID SIEGEL: OK, but, again-- we're actually talking--
ANDREW MCAFEE: There's a difference-- I think we can all agree that having both meaningful work and a strong safety net would be the ideal condition. A question that I don't know the answer to is, what if you have that safety net, but work goes away at a very large scale?
DAVID SIEGEL: And I can comment--
ANDREW MCAFEE: I still worry about that.
DAVID SIEGEL: And I can comment on the European economy. And so, again, we're not talking about the today [AUDIO OUT] point in time. So the question is, as this trend continues-- say, for 20, 30 more years-- what will the picture be? And what are the policies that are needed? Because enacting these changes takes a considerable amount of time.
Now I know a little bit about this from my own profession. You know, it is absolutely not the case that Europe has solved these problems. In particular, arguably, Europe is bankrupting itself in its economic policy. And so, at the moment, the situation in Europe, by some measures-- from essentially societal fairness and happiness-- is better. But by many measures it's completely unsustainable.
ANDREW MCAFEE: You wanted to jump in here.
ANDREW MCAFEE: Hold on. Hold on. Someone's been trying to get in here for a while.
DAVID SIEGEL: I would agree, for sure. But I think that the issue that-- I'm confident that the issue that we have is the transition problem. So we are in one state and we need [AUDIO OUT] to a different state. Now that transition-- and so, first of all, we don't know--
ANDREW MCAFEE: Do you share my belief that we're not working hard enough on that transition problem?
DAVID SIEGEL: Precisely.
ANDREW MCAFEE: OK.
DAVID SIEGEL: And so there are-- I am an optimist. And so I'm very in agreement with most of the views being given here that there is another state that will work for our society. But there may not be, first of all, agreement on what that state is. So step number one, we need to, first of all, estimate how quickly automation in AI will push us into that state.
And, number two, we have to get some kind of agreement on what that state should be. Number three, we have to do the economic analysis to make sure that that state is stable and not a mythical dream. And then, number four, we have to gradually-- we have to find a path to that state that does not involve war or famine or other bad things.
Because, [AUDIO OUT] often in our world, these path-state changes involve millions of people dying and complete restructuring of society. We all know the history of modern times.
ANDREW MCAFEE: Yeah?
AUDIENCE: Who's we?
DAVID SIEGEL: We are-- for example, you know, [AUDIO OUT] MIT.
DAVID SIEGEL: No. I mean, basically, this is a research problem. You know, the answer is academia, government, you know, it's a collective problem. There's no--
DAVID SIEGEL: I think that the United States-- we have, actually, a fairly self-contained economy, so that the trade issues and other things are relatively minor. Mostly because it's a heavily service-oriented economy to begin with, we have the benefit of being able to work on this problem without having to take into account the entire planet Earth. Would you agree with that?
ANDREW MCAFEE: Yeah, absolutely. But, Mark-- also to your question-- my favorite unit of analysis for thinking about different approaches here is the good, old-fashioned country. I think we'll see very different approaches country by country, as opposed to region by region, as opposed to either global region or within a country region.
DAVID SIEGEL: But if your country were smaller or more dependent upon, essentially, globalization, then the problem is actually harder, because then you need more core-- so this is the problem in Europe, because Europe is some hybrid thing, which will interfere with state change.
ANDREW MCAFEE: One of the dismaying things to me is that it's not just the workforce that is getting polarized in America. The evidence is very [AUDIO OUT] political parties are moving steadily farther apart. And our ability, like you say, to come together and make some big decisions-- that ability is decreasing instead of increasing.
And when you look at previous big technology transitions in the economy, or in other economies, what you see is that the country came together and made some pretty big [AUDIO OUT] My favorite example is universal primary education in America. Big decision, extremely controversial at the time, turns out to have been the right thing to do.
But I don't even see us coming together to try to collectively figure out what the big decisions are. There's way too much of, if you say it, I disagree, because you're blue and I'm red.
DAVID SIEGEL: But there are-- I agree completely-- but there are also small decisions people can make as well. And so when, for example, you know MIT students are thinking about companies to found and start, more people can be encouraged in their education to think about activities that might benefit from the use of human labor, rather than activities that are trying to automate away and reduce.
ANDREW MCAFEE: A tiny bit of self-promotion exactly on that front-- Erik Brynjolfsson-- Tommy, that's how to pronounce his name-- and the rest of us at the Initiative on the Digital Economy, which [AUDIO OUT] school, we've launched a thing called the Inclusive Innovation Challenge, where we're exactly trying to highlight the entrepreneurs and the innovators who are creating economic opportunity for good, old-fashioned, middle American people via tech progress.
DAVID SIEGEL: That's right. And one thing that I find interesting when I talk about this, people [AUDIO OUT] give me 10 ideas for companies that would benefit from increased-- and I'll say, well, wait a minute, why do you think I would know this?
ANDREW MCAFEE: Right. Let's crowd-source that.
DAVID SIEGEL: Let's crowdsource it, just like we do all inventive activities. So don't expect me to have the answer.
ANDREW MCAFEE: The previous history of universal basic income experiments-- in America, anyway-- is not an encouraging one. We did a couple studies, a couple efforts in Seattle and Denver in the '60s and '70s and--
ANDREW MCAFEE: Yes, no-questions-asked, basic income to a set of families and then didn't do it for another set of families.
ANDREW MCAFEE: I forget. It was a government effort. I don't know if it was state or federal level, but it absolutely happened. And the results were that people [AUDIO OUT] those got that money, they worked fewer hours and they-- I'm oversimplifying-- they withdrew from the labor force even more, which I think is the wrong direction. And their marriages fell apart at a higher rate, which I also think I think is the wrong direction.
DAVID SIEGEL: I think part of the problem here is the push and the pull. And, again, this, I think, points out some policymaker issues. So the employers like to say that the problem is that we don't have enough trained workers, we don't have people who are in-- right? But the workers, then, you know, even if they get retrained, can't find the employers.
And so there's a bit of finger pointing here. But I think the [AUDIO OUT] does, I think, show-- is it a labor-demand problem or a labor-supply problem.
DAVID SIEGEL: By the way, I think you're making an excellent point. And this is not something that we've talked about, which is-- and this is kind of unrelated, in a way, to AI and automation-- it's just the pace of change in our society has rapidly accelerated.
It took-- I mean, I like this analogy-- it took something like half a generation or more before the telephone became integrated into a routine human activity. And so that gave businesses-- everyone-- pretty much their working lifetime to adjust to the impact of being able to make a telephone call, which is great.
Now in our new society, you're expected-- because, you know, we're very innovative, and things are changing very rapidly-- to constantly be able to incorporate these new technologies and ideas into traditional activities.
This becomes overwhelming for people. So one reason, I believe-- no data on this-- that you have a withdrawal from the workforce is, we're just exhausted. It's like, I don't want to get more retraining. I can't do it. It's like, every time I get retrained, you tell me I have to retrain for something else. I mean, there's evidence that, essentially, people are just simply discouraged that they can't keep up.
Yeah, well, you know, I'll make an analogy. So if you install a robot in your factory, you do that with care. And you're not going to throw it away the next year because it was an investment-- not only the cost of the robot but, you know, programming it, fitting it into the process.
So you're going to cherish your robot and use it for its lifetime. And then you'll retire it and then I actually buy them on eBay and I have a collection of these service robots that, you know, did good work in factories. And so the robots-- [AUDIO OUT] the weird thing-- we're like giving robots careers. But with people, it's like, you can come and go. If business trails off a little bit, I'll lay you off. You know, who's going to take their robot and put it out on the street. But you put--
So there is actually-- so what I mean by jobs versus careers is a job is, basically, very transactional, that you would basically-- you know, it's employment at will. It's more of the American way versus the European way, I guess, or the Japanese way.
But a career is where you're going to nurture-- like if you have a robot, you know, you're going to have someone taking good care of it, oiling it every day, improving the programming to make it more efficient.
So these days, oddly, we're developing a much more intimate relationship and caring relationship for our automation and much-- you know, if an employee has a bug-- [AUDIO OUT] you fire the employee. If your software has a bug, you very carefully fix it.
AUDIENCE: Well, you know, there's another edge to that story.
DAVID SIEGEL: Right, Mark, I think that you may be drawing too much into the particular circumstances of your business. So another problem in our economy is that there are a very small number of job skills that are in high demand. And that's sort of the stuff that you do. [AUDIO OUT] --anything involving machine learning, big data.
So because everyone wants to hire these people, and there just isn't a big enough supply, you have that problem. But in a much more balanced economy, I think that workers are much more likely to be loyal because they're not-- [AUDIO OUT] they're quitting because they're getting phone calls from 10 companies trying to hire them because there aren't enough trained workers.
DAVID SIEGEL: Well, look, I mean, I think that the turnover rate that you have in a company is the result of a bunch of factors, not just worker disloyalty. I mean, it's a complicated question. And it has a lot to do with weight-- I mean, look, we've studied this problem pretty carefully.
And, usually, you have higher turnover, the data would show, in an industry that is having rapid wage growth. Because what happens is that it's very hard for you to get compensation correct in an organization if you're essentially having quick wage escalation. You don't want to completely readjust everyone. And it leads to unfairness and so on. People get larger offers.
And so the real turnover problem generally occurs in business areas where wages are growing quickly. And that's just, usually-- I mean, as your data shows-- that's not a particularly big part of the economy.
ANDREW MCAFEE: Economists tend to like churn. They tend to like people changing jobs. They tend to like labor force flexibility and fluidity a great deal. My economist colleagues would actually get nervous if we started to see people only having 30, 40-year stable careers. That would probably be-- we'd take that as a leading indicator that things were heading in the wrong direction.
DAVID SIEGEL: But back to my-- I didn't really answer, Josh, your question-- so what I meant by career-- and, again, I think that this actually reduces [AUDIO OUT] --over. At least, that's been my own experience-- that if employees feel like they have a pathway to wage growth and skill growth in the organization, then that, I think, you know is-- first of all, it's good investment in human beings.
But, secondly, it [AUDIO OUT] lead to lower turnover. We survey all employees that are exiting our company and, you know, one of the top reasons that-- we don't have very much turnover-- but one of the top reasons given is like new opportunity elsewhere. It's something that they can't do-- in other words, [AUDIO OUT] would be, put another way, maybe stagnation.
ANDREW MCAFEE: Yeah. Reid Hoffman, who's a LinkedIn founder and very good venture capitalist, has a nice way to think about that. He says we should maybe think about careers as consisting of successive tours of duty. And you and your employer should have a handshake agreement when you start work about what's the tour of duty [AUDIO OUT] how long is it going to be, which each side's responsibility.
And, at the end of that, if you want to sign up together for another tour of duty, go do that. I kind of like that as a mental way to approach that, as opposed to saying, wow, we should have long careers or we should have tons of short churn. This notion of a tour of duty is helpful to me.
DAVID SIEGEL: And now I think that if you have wage growth-- so if you can offer wage growth to employees, I mean, that's a good thing.
ANDREW MCAFEE: Wage growth solves so many problems.
DAVID SIEGEL: So you could almost define career-- giving someone a career-- as giving them the opportunity to have wage growth. And so then you have to be [AUDIO OUT] your business to see how you can design your company and do that.
ANDREW MCAFEE: Go ahead.
ANDREW MCAFEE: Despite having written two books, we're still not clear. Let me try to say a couple of things about that. You're right that [AUDIO OUT] between the two books that Eric and I wrote changed. And the reason for that is that we think the lead is that technological progress is a profoundly good thing.
And in The Second Machine Age, we wanted to tell that story first and try to communicate the bounty that comes [AUDIO OUT] very powerful technologies. Because I get way too many calls from journalists who want to write the robots are eating our jobs and, therefore, aren't the robots terrifying story. I think that's just getting it deeply wrong.
Economists will admit that they blew the argument on trade and, therefore, protectionism is back. It's failing Econ 101. It's a deeply terrible idea. But because the smart people didn't find a way to articulate the arguments, now we're talking about trade barriers and stuff like that.
So the reason that we tried to lead with the case for optimism and for liking tech is because that's the big story. The things that we've been talking about here are [AUDIO OUT] important consequence, I believe, of technological progress. And we need to talk about that along the lines that we've been doing this afternoon.
Now, to your question about education, it's a phenomen-- it's a great question. I think deeply rethinking education is a necessary condition, but not a sufficient one. What we need to do is, I think, two-fold. We need to get faster economic growth. Because faster economic growth, like we said earlier, solves so many other problems. Right? And so doing that is an urgent thing.
The other thing we need to do-- and this gets to some of our comments earlier-- and I'll be a little parochial and talk about America-- we need to rethink our safety [AUDIO OUT] all messed up. And your health benefits, your retirement benefits, all of these things are way too tied to whether or not you have a classic, industrial-era job. It's just very, very inappropriate.
In addition to which, Eric, my co-author, has a really nice way to say it. We are not anywhere near peak labor. Nowhere near peak labor. Bob, we might get there in the next 66 years. I think that we will. But we're not there now. And there is no shortage--
AUDIENCE: What do you mean by peak labor?
ANDREW MCAFEE: Peak labor would be when, to grow the economy, the number of hours worked has plateaued and is now going down year-by-year. We are not near that point, by no stretch of the imagination. The number of hours worked grows in lockstep with the growth of the overall economy.
When those two lines decouple, then we're in a very different era of our history. We ain't there yet. And there's no shortage of additional things that need to be done. We need to be educating people better-- the way we take care of our elderly, I think, is a national-- is a moral scandal. The number of elderly Americans who don't talk to another human being for days on end is staggeringly large.
We've got to take care of these people who are killing themselves quickly and slowly. There is no shortage of work to be done. Will market forces and entrepreneurs do all that work for us? I'm an ardent capitalist. I don't believe they will. Larry Summers has a great phrase-- he says there's not really a great business model for educating the children of very poor people.
So these are things that we can take on as a society and absolutely do. Let me say one more thing. And we are wealthy enough as a society to start doing this work that needs to be done. And I don't think that takes us down the road to crazy socialism, like some of my more libertarian friends think.
So, for all those reasons, if we did this hard work and got the policy environment right, a lot of these things that I was talking about-- I don't know if they'd go away-- I'm pretty confident they would be eased a [AUDIO OUT] And we have enough time to figure out what a meaningful life and a healthy community look like when they are no longer dominated by work.
I think we're going to see that-- I'll see that world before I die. We're not going to see it in the next decade-plus. I do not believe that, even with Mark's five-year horizon [AUDIO OUT] truck unloading robot.
DAVID SIEGEL: Yeah. And to add to your point, there's all this talk about the need for improved infrastructure in the country. The kind of work involved with improving the infrastructure, I believe, Mark, is going to be relatively hard to automate in the next 10 years.
ANDREW MCAFEE: And it's exactly the kind of work that would get these stereotypical brawny American men off the sidelines.
DAVID SIEGEL: But here we go back to the structure of the economy question. So it would seem sensible to me that we should continue to be importing all these great [AUDIO OUT] at low cost from China and other countries and take, you know, essentially these workers and get them involved with infrastructure.
So why doesn't that happen? See, the way we rigged up our economy is that there is no revenue model for most infrastructure. So, for example, replacing sewers, building roads, and so on, by and large, it doesn't generate revenue the way when a company sells a product they can get people to directly pay for it.
And so what we have is, again, just because of the way we've rigged up our incentive system, we have disincentive for employing people in building [AUDIO OUT]
ANDREW MCAFEE: You have to go way outside of mainstream economists-- you have to go way to the left or way to the right before you'll find any decent economist who doesn't believe that government should be paying for sewers and infrastructure like that. Those are fringe weirdos who don't believe that.
ANDREW MCAFEE: Yeah, it's called--
DAVID SIEGEL: Yeah.
ANDREW MCAFEE: Yeah it's called tax.
DAVID SIEGEL: This is why I think we're-- I mean, we can go on all night-- but I think one of the reasons we're killing ourselves here is that the things that we really need, that would employ a lot of workers, we don't want to have a revenue model for it. And the only way we want to pay for them is with taxes.
But, at the same time, people don't want to pay taxes. They want to have their money to buy, you know, other things. And so this seems like a diabolical policy bind here.
ANDREW MCAFEE: Be careful. [AUDIO OUT] basically, most other rich countries don't have this particular bind. This is a particularly weird American--
DAVID SIEGEL: Sorry, I'm talking about America.
ANDREW MCAFEE: I understand that. But that's not any insoluble policy bind.
DAVID SIEGEL: I agree.
ANDREW MCAFEE: It's the fact that too many people don't think that infrastructure is a thing, evidently.
DAVID SIEGEL: I totally agree.
ANDREW MCAFEE: But our corporate tax rates-- they're high on paper and [AUDIO OUT] reality. It's a full employment bill for really, really good accountants. So, actually, getting companies to pay their fair share would be massively revenue generating. Use that to improve our bridges and put people to work and then we have better bridges.
Look, this is not even Econ 101, right? Alan Blinder is a really good economist [AUDIO OUT] -is that the areas where economists agree the most are where they're listened to the least. And infrastructure is exhibit A for that.
My headline is don't stop innovating, right? Tech--
ANDREW MCAFEE: OK. I just want to say, because very often people say, look, the innovators need to [AUDIO OUT] the pace of what they're doing so we have a chance to catch our breath. I don't believe that.
AUDIENCE: I have this dream-- [INAUDIBLE] How do we get there?
DAVID SIEGEL: Well, a simple suggestion that I would make, which, I think, is easy is that the people here at MIT should spend some extra time trying to realistically communicate to the world-- not in academic journal, [AUDIO OUT]
ANDREW MCAFEE: That's right.
DAVID SIEGEL: --about what to expect. And, you know, not the Kurzweil way of doing it--
ANDREW MCAFEE: That's right.
DAVID SIEGEL: --because that just is a fog and confusing. But level-headed-- just the way you're doing it-- but I think there can be a much broader dialogue in society to raise awareness of what's happening in a [AUDIO OUT] realistic way.
ANDREW MCAFEE: If you all make the same mistake economists make, you've got a very serious problem, and shame on you. Because economists have shown what not to do. They just lost the debate on trade.
If the technologists lose the debate on tech progress, we have two very serious problems instead of just one.
ANDREW MCAFEE: It's tough because we don't just need to keep talking to each other.
ANDREW MCAFEE: I think that's a failure mode that's very easy to get into and I thinked we've [AUDIO OUT] dangerous that can be.
DAVID SIEGEL: Well, I mean, look, I think that little op-ed I wrote in The New York Times, I was surprised at how much attention it generated, when I considered it to be actually, you know, really just stating the obvious.
ANDREW MCAFEE: Tommy, I would say both advancing and broadening the conversation and working really hard on both those things is important. If all of us, you know, people at elite academic institutions just keep talking to each other, massively insufficient.
DAVID SIEGEL: Look, and the other thing I would add, just to repeat what I stated-- and this is more of an MIT thing or a broader university throughout America-- is that I think that there should be more emphasis, particularly in entrepreneurialism, to try to get people solving a more diverse set of problems. I think it's become too narrowly focused.
ANDREW MCAFEE: And we're celebrating Narrow A Thing, which is the next app, or the next Snapchat, right? Let's broaden out that celebration.
DAVID SIEGEL: You know, just before-- just one thing I like to point out about how this works-- one example I like to give is like a Facebook. So, you know, Facebook-- you know, everyone in America uses it-- mostly everyone in the world. And people, I think, in America use it for an average of 20 minutes a day, maybe more.
And this company is only employing, you know, 25,000 people to serve this global community of three or four billion users. So we can kind of round down the number of people they're employing to zero. And generating-- consuming-- 20 minutes-- I think it's even more-- on average, of our time.
And so those are 20 minutes that we can't consume other activities. So, essentially imagine, if everyone is focusing on this Facebook app, to try to get people-- billions of people-- to spend 20 minutes a day using it. I mean, maybe you have 10 of these apps and eventually all we're doing is-- and then you'll have a collapse in economic activity.
ANDREW MCAFEE: One of the scary things-- one of the other scary things in the economy-- entrepreneurship is on the decline in America. And Silicon Valley distracts us. Because it's so bright and shiny and alluring. Good, old-fashioned entrepreneurship is on the decline.
And the reason that's really scary is that young, high-growth companies have been one of the big job creation and wage growth engines in the economy. Here's the scarier part-- we don't understand very well why entrepreneur [AUDIO OUT] decline. It feels like death by 1,000 cuts, but we actually don't understand it very well.
ANDREW MCAFEE: There are so many different ways to look at it. Any measure you want to capture of new venture formation, new establishment creation, new company creation, percentage of Americans employed by young companies-- they're all on a long, slow, steady, 20-year decline.
And figuring out how to reverse that is-- more research is needed is kind of a stock phrase. More research is desperately needed here. You've been super patient. Thank you all.