The emergence of Big Data, the exponential growth unleashed by decades of Moore’s Law (more and more computing power for less and less cost), and the logic of what the authors call “recombinant innovation” — the mixing and matching of our powerful new tools into a bewildering array of even newer, even more powerful tools — have replaced hype with a bewildering new reality. We’re headed somewhere new at high speed, and with no apparent ability to put on the brakes.
“The Second Machine Age” is fascinating because it is simultaneously hopeful and wary about how technological change is remaking our lives. The authors make a compelling case that the second machine age will deliver vast “bounty” to humankind. The overall size of the economic pie is sure to grow. As consumers, the options available to us will beggar description. But McAfee and Brynjolfsson are also quite clear-eyed about the alarming reality of how that pie is being sliced up and distributed. The numbers can’t be ignored: The bounty is growing, but so is what the authors call “the spread” — growing income inequality, greater concentration of wealth in fewer hands, unprecedented pressure on labor markets.
How did this happen, and what should we do about it? Brynjolfsson and McAfee spoke with Salon by phone to discuss the challenges, and opportunities, of the second machine age.
You write: “The Second Machine Age will make mockery of all that comes before.” What does that mean? What is the second machine age?
Andy: The first machine age is the Industrial Revolution. The first machine age was essentially about amplifying our muscle power. What Erik and I are saying in “The Second Machine Age” is that we are now in the early stages of a parallel acceleration and amplification of our cognitive abilities. The line you quoted is an allusion to something Ian Morris said in his fantastic book, “Why the West Rules for Now.” As Morris was graphing human history, his lines went from basically horizontal to vertical at the outset of the Industrial Revolution. “It made mockery out of everything that had come before”: wars and empires and everything else.
Erik: These titanic changes have had a huge effect on human living standards, and in many ways we can learn from how things changed with the first Industrial Revolution and the first machine age. But there are also some key differences. As we augmented and automated muscle power and our ability to use that power to manipulate the world — not just with the steam engine but with subsequent general-purpose technologies like the internal combustion engine and electricity — it acted largely as a complement to human decision making. The power wasn’t very valuable unless you had someone controlling it and deciding what to do with it.
With the second machine age we are augmenting and automating a lot of cognitive tasks and it is not as clear that humans will be a complement to those kinds of machines. In some cases the machines will be substituting for humans. And that has a different kind of effect on the labor markets as well as on economic output.
So what, exactly, happened? For decades people have been making extravagant claims for the transformations that were sure to come from networks and digital computing. But for most of that period hype outpaced reality. Now, in the last four or five years, thing really seem to be speeding up. What changed?
Erik: We need to keep some historical perspective here. The early revolutions took a century or more to play out as each new general purpose technology kicked in. So we still have a lot ahead of us. I think we still have a century or more of complementary innovations to play out. But specifically to your point of what has happened recently — Andy and I have been astonished at how quickly things have just started happening in the past five or 10 years.
We were teaching a course where we talked about what machines could do well and what humans could do well, and we gave fine motor control and pattern recognition, and specifically driving a car, as examples of things that humans were particularly adept at and we didn’t think that machines would do any time soon.
Six years later we were riding down Route 101 in one of Google’s driverless cars! So we were wrong in how long that would take.
A big part of that has been the big data revolution, the ability to get around or sidestep some thorny problems that researchers have been working on for decades in language recognition and the perceivable world, not by using the traditional methods, but really just by throwing huge amounts of data at the problem. That’s one of the three big forces changing the world right now, along with exponential growth and what we call combinatorial growth.
Andy: There’s a great quote from Hemingway about how a man goes broke. “It’s gradually … and then suddenly.” And that really characterizes how digital technical progress has unfolded. Like you say, people have been making these big claims about what computers and robots were going to be doing for about half a century. And now we’re kind of there. That’s what motivated us to write the book. We were like, wait a minute: We didn’t think our cars were going to drive ourselves or our phones would talk to us without a person at the other end of the line. The exponential part of the story has to do with the fact that after Moore’s Law has been going on long enough, a difference in degree does become a difference in kind. And finally there is this idea of innovation as a combinatorial process. What people are doing is taking all these previous building blocks and adding digital blocks to them, and that’s led to this growth spurt that we’ve been seeing.
The other thing that we’ve been stressing is that both of us believe we ain’t seen nothing yet. What people are going to continue to do with these tools is just going to blow us away.
But is that something to be excited about or terrified of? Isn’t it a staple of science fiction that the kind of multi-recombinant exponential growth that you describe generally leads to Skynet or the Matrix?
Erik: Well it depends on which kind of science fiction you watch. There are at least two big genres out there. There’s “Star Trek” as well. But you’re right. We see two camps. There are the dystopians who see really bad outcomes. A lot of economists point to the relative stagnation of median income as a warning sign. But there is also some really good news. Innovation creates wealth. We are at record levels of wealth in this country, we are up to $77 trillion in household wealth. That’s a new all-time high. That was also part of what motivated Andy and me to work on this; this confusion we had about how you could have so many good things happening and so many bad things happening simultaneously. After digging into the data and talking to people we concluded that both the good and bad events had a common cause — this technological digitization of the economy.
The key economic fact is that technology can make the pie bigger, it can make the economy richer, but it doesn’t necessarily help everybody, in fact some people, even the majority of people, can be made worse off, even if the pie grows bigger.
That’s not what happened for the past 200 years, with the first machine age, but it does seem to be what’s happening now. In the second machine age we have a bigger pie but also more concentration of wealth. The median income is now lower than it was in the 1990s.
Andy: There’s no law that says technology makes everybody better off. Some people appear to be worse off in their role as people who want to offer their labor to an employer or to the economy. But the point about the pie growing is still a really important point, because in our roles as citizens or consumers, technological progress is fantastic news. Even as the music industry has shrunk we’re all listening to more music. Warren Buffett can’t buy more Wikipedia than anyone else can buy. Our health outcomes are improving a great deal.
You brought up Skynet. As Erik and I looked around we found ourselves becoming less worried about machines becoming self-aware and rising up against us, but we found ourselves more concerned about how, in our roles as workers, things are really getting challenging for a lot of people.
That does seem to be the defining question of the moment.
Erik: But none of these futures that we describe as possible are inevitable. You can have some really bad outcomes, you can have some really good outcomes — it is all going to depend on how we respond. The technology is only going to accelerate. If we respond correctly, then this is going to be good, good news. But it’s not going to happen automatically.
Whenever we talk about technology and jobs, some economists, and certainly lots of voices on the more conservative side of the political spectrum, will argue that any jobs that get lost in one sector of the economy will be more than compensated for by new job creation in other sectors. But that assumption seems to breaking down.
Erik and Andy: Yup!
Andy: There seem to be two things going on. Number one is that those commentators that you mention are looking back at the historical pattern and getting a great deal of confidence from it — as they should. The historical pattern is one of a succession of pretty smart people saying large-scale technological unemployment is right around the corner, and basically being wrong. The question for us, and Erik and I spend a lot of time on this: Is this time different?
What those commentators get wrong is the fact that while technological progress does grow the pie, there is honestly no economic law that says that growth is going to float all boats the same way equally.
Erik: This is the big question: whether or not the jobs will be there going forward. In the past they always have been, but Andy and I don’t think that’s automatic. Technology has always been destroying jobs, and it’s always been creating jobs, and it’s been roughly a wash for the last 200 years. But starting in the 1990s the employment to population ration really started plummeting and it’s now fallen off a cliff and not getting back up. We think that it should be the focus of policymakers right now to figure out how to address that, because it is likely that technology is going to have an even bigger impact going forward. So we can’t just ignore it.
So what can we do?
Erik: There are three high-level categories: education, entrepreneurship and tax policy. Each of those things needs to be reinvented and rethought. We need to fundamentally reinvent education, just like media and publishing and retailing and manufacturing and finance and just about every other industry has been reinvented. Our industry has been a real laggard. We need to rethink the philosophy of having people sit in rows and learn how to follow instructions; we should be fostering creativity, because the rote kind of skills are exactly what’s being automated. On tax policy, instead of punishing people hiring workers we should be rewarding them. Right now we tax labor, and one of the basic laws of economics is that if you tax things you get less of them. That wasn’t much of a problem for much of the previous history, but now it is having some really negative effects.
Andy: And the only thing I want to add on to that is that in the short term the robots and the androids and the AIs are not about to take all of our jobs in the next month or the next year. The progress, while astonishing, is just not that fast. So the right thing to do today is what Erik and I call the Econ 101 playbook for stimulating economic growth: education, entrepreneurship, infrastructure, immigration, basic research — you are not going to get disagreement among well-trained economists on these kinds of things.
OK — but the Econ 101 playbook is the standard prescription for how to spur growth regardless of the external circumstances. Good times or bad, it’s what most economists would advocate. But you make a very strong case in your book that current “advances in technology are driving an unprecedented reallocation of wealth and income.” If we’re going through an unprecedented transformation, is the standard playbook going to be enough to cope with it, or are we going to have to explore more drastic alternatives?
Erik: Economic laws haven’t changed. We are in a different part of the technological landscape than we have been in the past, but basic principles like, if you tax something you get less of it, and if you don’t, you get more of it, are still true. We don’t have to throw out the concept of economics to understand how to respond to it the current situation.
Andy: For example, if we were to upgrade this country’s infrastructure so that the civil engineering society would give us a decent grade, instead of a D+, that would set the table for a much better business climate in the United States. Robots are still very bad at repairing bridges, so spending on infrastructure would result in jobs. If we could fix our immigration policy and let in these incredibly skilled people, talented people that want desperately to come to our country, if we could do that, we know they would create jobs. Immigration is a huge vehicle for entrepreneurship in this country. And if we make the climate for starting a new business more favorable we know the job growth would come. Over the next decent chunk of time we’re still confident that the Econ 101 playbook is the right answer even as the technology continues to race ahead. If that’s anywhere near correct, that gives us time to think about what to do, if we really are heading into this sci-fi future down the road, and we both think that we are.
But how do you deal with the political consequences of growing inequality — the broadening “spread.” You point out that the people who are benefiting the most from the current technologically driven productivity growth are the owners of capital. You can make a good case that as a result of this vast increase in wealth, the owners of capital have more political power now than they’ve enjoyed in at least a century. I would argue that they are using that power to stymie the kind of appropriate policy responses that would reduce growing spread and reverse spiraling income inequality. Technological change isn’t just screwing with the job market, it’s concentrating wealth in the hands of people who are actively resisting any efforts to ameliorate the problem. So the spread has serious political consequences. That seems like a really hard problem to crack.
Erik: It is. And we are very concerned about it. Our colleague Daron Acemoglu co-authored a book called “Why Nations Fail,” where he goes into great depth about how economic concentration can lead to political concentration exactly along the lines that you are describing and how that can lead to a vicious, self-defeating cycle that brings down the whole economy. We certainly don’t want to get into that kind of cycle, and that’s one of the reason we want to change the conversation to focus more on understanding what seems to be driving these changes in the economy. If we get the diagnosis right, I think we would be in a better position to get the right prescriptions.
Andy: What we don’t think is the right prescription is to say let’s make sure nobody else ever gets rich off new technology. That’s really not the way we want to go ahead.
Erik: But it has to be balanced with an equality of opportunity.
Andy: And the big worry, the reason why we quoted Daron in “Why Nations Fail,” is that he makes a very persuasive case that the rising inequality that you describe will eventually lead to a serious decrease in the quality of opportunity. And if that’s under threat we need to be concerned about that.
But how do you break the cycle when the logic of technological change is contributing to that cycle?
Erik: Well, you can’t take the attitude that there is nothing we can do. We do believe that technology is a driver of change, but it doesn’t follow that there is nothing we can do to shape the future. We think that education, entrepreneurship and tax policy can all help in increasing both the bounty and decreasing the spread. That’s our grand challenge.