Discussion: (0 comments)
There are no comments available.
A public policy blog from AEI
According to MIT’s Andrew McAfee, most of us vastly underestimate the pace of technological innovation. What’s more, this technological revolution will be as transformative as the Industrial Revolution before it. He recently joined me on the podcast to discuss the ramifications of this change for businesses, the economy, and society at large.
Andrew McAfee is a Principal Research Scientist at MIT where he studies how technology is changing business, the economy, and society. He is the co-author of several books with his colleague Erik Brynjolfsson, including “Race Against The Machine,” and “The Second Machine Age,” and his most recent one, “Machine, Platform, Crowd: Harnessing Our Digital Future.”
“Machine, Platform, Crowd.” Your initial example for each of those is, for machine, AlphaGo besting the best of human players at Go; platform, Airbnb upending the hotel business, with none of the traditional assets (they don’t own a bunch of hotels); and then you gave an example of General Electric using an online crowd to help design its market products.
So, you have machine, platform, crowd, and all those have a counterpart, a kind of pre-Digital Age economy counterpart. For machine, you have the human mind; platform, you have companies making goods and services and products; and crowd, rather than crowdsourcing things, you just do it internally. And so the thesis of your book is that there needs to be a rebalancing toward the machine, platform, and crowd from these other things.
And away from an over-emphasis, we believe, on minds, products, and the core of the organization.
So, first of all, maybe the easy question is: Why does it need to be done now? Or why should this be happening now?
Because during periods when there is a great technology surge, organizations need to rethink their business models; the smart approach changes. And the most recent example we have of this is the transition from steam power to electric power. And the smart way to build and run a factory in the era of steam was not a little bit different than the smart way to build and run a factory in the era of electricity; it was night and day. And companies that weren’t aware of this and just thought, “Oh, here comes an electric motor, we can replace that big steam engine in our basement with a big electric motor,” and then stopped there — they just got swamped by competitors who came along and said, “You know what electricity actually lets us do? Conveyor belts, overhead cranes, assembly lines, machine shops, a completely different way to think about the process of making things in a factory.”
And the incumbents, largely because of their mindsets and where they came from and their past successes, did not get the new possibilities. And what my co-author Erik Brynjolfsson and I are trying to do with this book is essentially say to the incumbent economy that there is a change happening that is as big as the transition from steam over to electric. And if you’re thinking that what’s happening is the ability to run your steam-powered factory a little bit better, you are making an equally big mental error.
So, this is a book meant to enlighten and make aware big incumbent companies of these changes. But can they adapt to them? I mean, hasn’t the story been that the big companies are just unable to be nimble enough, to change their corporate culture, to adapt to these new technologies? Do you have any confidence that even if a CEO reads your book and takes it to heart that they’ll be able to adapt?
Yeah, I think that’s a great, open question. And like you point out, history does not make us confident. Incumbent organizations, over and over again, miss disruptive change. And they find their margins eroding, they find their customers going elsewhere. They kind of sit around and go “what the heck happened here.”
The grounds for optimism that I have are essentially that smart organizations are aware of that phenomenon. And because of the work of Clay Christensen and a lot of other people, I think that executives and companies today are more keenly aware of the possibility of disruption, and they are watching it happen in front them in lots of different industries. So, they do have awareness, and they have got fire in their bellies.
The huge open question is even with that awareness at the top, which I think is a necessary but not a sufficient condition, can big successful incumbent companies change themselves? Like you point out, these are large organizations, they’ve got their norms, they’ve got their culture; they are hard to change. And the image that people use, which I really like, is of a tugboat trying to change the direction of an ocean liner. I don’t know how quickly some of these ocean liners are going to be able to turn, and it’s not like the disruptors are giving them a lot of slack.
And so, we’ve got a really interesting competitive battleground going forward in industry after industry. Both the industries that you and I would already know about — everything from music to journalism — to groceries now, to giving rides across town in a car. This disruption of the technology surge are manifesting themselves in pretty weird corners of the business world. Ones that I would have expected, and ones that have caught a lot of people by surprise.
Is there a classic recent example of a company making this adjustment to the new digital economy? Like an older company, at least at this point, that seems to be getting it right?
One of my favorite recent examples — and this is the stuff that Clay Christensen did for his dissertation when he looked at companies in one specific industry — is the disk drive industry. He noticed that the leaders kept on missing the transitions from one kind of disk drive to another. He found a couple of exceptions to that rule, and if I am remembering right, the only thing that he could identify that they were doing differently was that their leaders were aware they were appropriately frightened about missing out and they pulled their organization into a new direction.
Let me give you a couple of examples where we have seen things like this playing out more recently. I am a big fan of Gary Loveman as a CEO of Harrah’s. Now that company recently had some trouble because of the Great Recession and because of big changes in the gaming industry. But for a long time before that, Gary came in and looked around at this kind of old-fashioned company in an old-fashioned industry, where you have got high-rollers and pit bosses and comping meals and stuff like that, and he said “No, what this actually is is an analytical engine, and we are getting huge amounts of data about the activities and preferences of all these people who walk into our door. Why are we not using that to give them very customized offers, to treat them with a great personalization, to segregate our customers to do kind of marketing 101, to segment your customers and treat them differently and try to get them to upgrade to the next level?” And so, he turned this company into an analytically-driven organization.
I think it’s a great example of, at least for a period of time, a successful transition to a more digital, more data-heavy future. More recently, I’m impressed by how one of the stalwarts of the industrial age, General Electric, is at least aware of the opportunities and the challenges facing that really successful organization. And I know what the leadership of the company has been doing a bit; they are enthusiastic and keenly aware. It remains to be seen how they are going to do against competitors, both old and new, in the different areas where they do business.
Let’s take a couple of points here. The machine and the human mind, and how they can work together: Now typically, when we talk about them working together, I think of machines doing the rote stuff, number crunching. And then humans do more interesting things. We show our judgment, we are the creative ones. And that’s how we’re going to survive the rise of the robots: doing that kind of thing. It was interesting when you were writing about this in the book, you said maybe not; maybe judgment and creativity are actually things robots and AI and algorithms can do. If they can do that, then what are we supposed to be doing? And why are we not as good as machines when it comes to judgment or creativity, or why are they getting better?
Yeah, and like you point out, we’ve walked around for a while, especially in the era of pretty powerful corporate information systems. Since we’ve had it for, call it 20 years, the division of labor we thought we had nailed. It was: Let the computer handle the rote stuff, the arithmetic, the transaction processing, and free up humans to do the stuff that humans are really good at. And just like you say, what we thought humans are really good at is judgment and intuition, and the kinds of stuff that only comes from experience, and knowing your environment, knowing your customer and trusting your gut and building up these intuitive, these judgment-based abilities over time. It turns out, that’s really bad advice, and it’s bad advice not because we have terrible judgment all the time, but because we have really really buggy judgment where we have all these cognitive biases.
When we were writing the book, we went to the Wikipedia page about human cognitive biases; there are about a hundred and seventy-five entries in it. Our mental computer has this huge set of bugs and glitches and flaws with it. Now it gets us through the day, but it doesn’t let us make optimal decisions in most cases. One of the most pernicious biases and problems with our wetware is that we are way too fond of our wetware. And we are unaware when we are making mistakes of how biased, flawed, and buggy we are. And we walk around patting ourselves on the back for how awesome our judgment is.
It’s hard to judge ourselves because we are also using the same mind.
Exactly. That mind is really really bad at turning any kind of rational spotlight on itself; it’s just another bias that we have. And, if you ask any roomful of people to have the audience raise their hands if they have below average judgment, you do not see many hands go up. We are all incredibly fond of our intuitive abilities and our judgment.
The data is overwhelmingly clear that it is really easy to make better decisions than even an experienced human expert will make by following the data and not applying any judgment on top of it. So, if you have a choice — and we described in the book lots of different studies, tons of different research — in most cases, if you have a choice between purely setting up even a simple algorithm, putting some data in it, and following what it says blindly, versus relying on the decision, the prediction, and the judgment of an experienced human expert, you will do better blindly following the data.
So that’s why I think that this balance between minds and machines needs to be rethought pretty substantially. But that doesn’t mean that minds are useless, because we don’t face a stark choice between only using minds or only using what machines spit out; we can combine the two. The good news there is that we make lots of errors, but we make very different kinds of errors than machines make. And if you can combine the two entities correctly, you don’t double down on your errors; the two kinds can cancel each other out.
And are there other skills — since it seems to be that we are narrowing the playing field as far as skills go — or are we all just going to end up as life coaches or business coaches persuading people to follow the decisions and judgments made by machines?
You say that in a little bit of a dismissive way, but we shouldn’t. If what we want is for better decisions a) to get made, and b) to get followed, then we need to bring together minds and machines and human abilities very differently than we do today. Medicine is my exhibit A for this. Medical diagnosis is a really difficult, subtle, pattern-matching activity. Machine learning systems are reaching superhuman performance over and over again, even in very difficult, subtle, data-rich, pattern-matching activities. That’s exactly what they are good at. So, I think the work of medical diagnosis will and should become a lot more machine heavy than it is today.
However, most people don’t want to get their diagnosis from an artificial voice, and most people are absolutely not going to follow their treatment plan. We know how big a deal medical non-compliance is and most people are not going to follow their treatment plan if it only gets spit out by a computer. So, the medical office of the future that I can envision includes a really compassionate, empathetic, like you say, a coach, about your health and your care path, combined with a whole lot of technology to make sure that the care path is the correct one. And I don’t find that a diminished future for human beings at all, I think it’s a very different future for the people involved in health care delivery.
The question you probably get a lot is what humans will do in this future. And I am wondering, as you talk to both regular people and people in the business world, do you think people generally underestimate or overestimate the pace of technological change?
Greatly underestimate it, and I don’t mean that to be dismissive of, you know, the average consumer of information. I underestimated it, Erik underestimated it, when we go talk to the alpha-geeks, the people who are building these technologies, they keep telling us that they underestimated it.
So, they did a survey a couple of years ago, about when there would be a computer champion in the Asian strategy game of Go. And the consensus was that was a long long way off; it’s an incredibly difficult, weird, pattern-matching exercise that we humans can’t even explain. The top Go players can’t explain why they made the move that they did. By all of our previous approaches to solving tough problems with computers, that puts us at a dead end. So most people, even insiders, thought that the computer Go champion was a long long way away. The world’s best Go player today is without question, hands down a piece of technology. So, we underestimated that transition. I see that crop up over and over again.
When you were talking earlier about how businesses are a bit more aware of how they can be disrupted by technology and new companies, I immediately thought of a very successful, somewhat new company: Facebook. There was an article recently about how Facebook tracks potential competitors and startups, and occasionally go out and buy them. And you see this in other places where companies buy a company — maybe they just want their technology, or maybe they actually see it as a potential challenge. And as I’m sure you are aware, these big platform companies have been subject of a lot of debate lately that they are all just going to become too big, moving into too many industries, and this is all part of a general worry about industry concentration. What do you think about that? Is there a concern that these companies are going to be too big, too powerful, too invasive, and that will be bad for the economy?
My ground rule is that all great concentrations of power demand vigilance. That’s different than having a ground rule that all great concentrations of power are bad and must be broken up. And when you talk specifically about these huge tech platforms, and their effect on competition and their effect on the economy — Erik Brynjolfsson, my co-author, my friend, uses this kind of a three-part test which I think is fantastic, so I am going to channel it. He says, first of all, what does the competitive landscape look like? In other words, do they make things a lot more difficult for competitors? And yeah, absolutely they do. I heard somebody else describe, a Silicon Valley insider, describe what a lot of startups are doing these days as, quote, “Free R&D” for Apple, Amazon, Facebook, Google, Microsoft.
In other words, they look over at what some startups are doing and say, “that’s a great idea, we’re just going to do it” and you know, because they have so many more resources and so much more reach, they can drive a lot of these startups out of business. Okay, we should keep our eye on that, but that’s only one of the three things to keep our eye on.
The second one is customer benefit, consumer welfare. And it’s really hard to argue that, for example, Google and Facebook are old-fashioned, price gouging monopolists, when they give a massive amount of stuff away for free to you and me.
Right, and I’m not sure what the number 3 is in that three-part test is, but I’m pretty sure these companies would like to focus just on the second part.
Which is fair though. I mean, are you and I going to get price gouged? Is Amazon price gouging us? Prices just seem to be going down with Amazon. And you can think that Bezos just sits up there in his lair, and waits till all of its competition is driven out, and then ratchets the prices on us, and inflicts misery on us, but we are nowhere near that point. So, it is very hard to argue that you and I as members of society are becoming worse off because of these big tech companies.
For sure, as far as the benefits they seem to be providing us, that is by far the strongest argument for saying that there is no negative argument.
Well here’s the third leg of the tripod that Erik brings up: How is the state and pace of innovation these days? Are we seeing a lot of it? Is it healthy? Or is it kind of anemic, being throttled by monopolists? Because the classic view of what monopolists do is either try to kill innovation, or they get lazy and don’t generate as many themselves, and I don’t see a lot of that. I struggle to keep on top of all of the fundamentally interesting and cool things happening in the technology space these days, a lot of them coming from the great big tech companies. So, of the three-part test — state of innovation, consumer benefit, and state of competition — we are 2 for 3 in the good column.
But you know the third one is controversial, and I think that has been the topic of my previous podcast with Eric. People will argue, well that’s great that you have some innovation that provides us benefits, but those aren’t really great benefits. And you can say that’s why growth is so low, and that’s why productivity growth is so low, and why you have an overall decline in the number of new businesses being started in this country. We do have a startup and innovation problem, and the example I gave you earlier about Facebook buying up smaller companies shows that maybe there is an issue, though maybe it is not reflected in the government’s statistics. So you’re confident that we remain a very innovative economy even if it’s not being caught in traditional metrics?
Yeah, absolutely, and I want to be careful to distinguish different measures of business dynamism, which you and I both agree, is an incredibly important thing for an economy. We want lots of activity, we want people to have more opportunities, we want more good stuff coming into our lives from the business sector. And like you point out, by a lot of measures, business dynamism is heading in the wrong direction. There’s less entrepreneurship, people are moving around less, companies in general are getting older, which is a weird phenomenon. But, when I think about another aspect of business dynamism, the state of innovation — is interesting, powerful helpful new stuff coming into the world? — I have a really hard time being a pessimist, I have a really hard time.
There was a piece in the New York Times — I think it was headlined “Breakup Google” — and the author Jonathan Taplin made the argument that these companies are too big, too powerful, they are squashing innovation. He didn’t have a three-part test but he had a three-part solution. His solutions were, first:
[T]hese companies should not be allowed to acquire other major firms, like Spotify or Snapchat.
The second alternative is to regulate a company like Google as a public utility, requiring it to license out patents, for a nominal fee, for its search algorithms, advertising exchanges and other key innovations.
The third alternative is to remove the “safe harbor” clause in the 1998 Digital Millennium Copyright Act, which allows companies like Facebook and Google’s YouTube to free ride on the content produced by others.
So the idea is that these companies need to be blocked from getting bigger, at least by acquisition, as well as being fairly heavily regulated. I think Mr. Taplin is probably on the left but I have heard people on the right making the same argument. What do you think of those ideas and do you think we need a lot more government regulation?
Yeah Jim, I think your career is like mine, in that it’s long enough to remember when people were making exactly the same arguments, for exactly the same reasons, about IBM in the early 1980s. They were making exactly the same arguments with exactly the same gloom and doom scenarios about Microsoft about 10 years later in the early 90s. We were worried about how dominant Netscape or AOL were going to be. Little more than 10 years ago, we were incredibly worried about the anemic state of the wireless communications industry in the United States, because it looked like Nokia was going to run away with everything. These all look incredibly shortsighted and quaint from the vantage point of 2017. I think we are going to look back from some future period and be equally incredulous that there was all this hand-wringing about this generation of great big powerful tech companies.
I love pointing to the example, I think it was Business Week of Fortune Magazine that had a cover about Yahoo! and their headline was “The search engine wars are over, Yahoo! has won.”
Right, so I have very little patience for these super-confident predictions that this big tech company is a monopoly, and an eternal monopoly, and therefore it’s really really bad, and therefore I know these really really big ways to intervene in the workings of the market: Force companies to license stuff, make them a utility. These are big draconian interventions in the market, and what’s going on in the present, and history, combine to tell me that they’re just really really bad ideas.
I think I tend to side with you on this, but just to take the other position, much of the work that you’ve been doing is to say this time is different, that for all the easy predictions in the past about technology and labor markets, technology is just different this time and we need to think harder about how we adjust, how we adapt, how we train workers, that we shouldn’t just assume that everyone’s going to be a winner over the long term. So maybe this time is different when it comes to Amazon, and Google, and Facebook as well, and they’re different than Yahoo, and AOL, and MySpace — especially given the amount of data they’re able to vacuum up and their ability to use that data. Maybe they are just a different kind of company than the ones that you were mentioning earlier.
Yeah, it’s a great retort to what I just said, so I have a two-part response to that. One is, again, my fundamental rule is that all great concentrations of power demand vigilance; so, let’s keep an eye on these companies, and see if they are doing things that we think are harmful to the outcomes that are valuable to us. And those include consumer benefit and innovation. And Jim you know the great quote that the point of antitrust policy is not to protect competitors, it is to protect competition. And I swap out competition sometimes and put innovation in there. The idea that any of these great tech companies, the big ones today, do not face competition — that’s a bad joke. We don’t even have to think too hard about that. Google might not face competition in search. But it faces intense competition in mobile platforms; there are a set of very very large and very very successful Chinese companies that have their eyes on global markets. So, the idea that any of these companies is a complacent, fat and happy monopoly that faces no threat — we don’t need to spend any time on that idea, it’s a joke.
I have written that if you look over the past 10 years, about the only thing that seems to have really gone right in the US economy is what’s happening in Silicon Valley. And the success of these big companies is providing amazing products and services, yet at the same time that’s where we are attacking right now.
So, by all means, let’s go smash them up, right, because some of us are uncomfortable with some aspects of that. The right answer is to make them utilities which are not the fonts of innovation in the economy, or to break them up, because we think that would be better somehow. I find some real arrogance behind those kinds of suggestions.
Some technologists are even contributing to that. They are talking a lot about technological unemployment. You have Bill Gates talking about robot taxes. There just seems to be a lot more apprehension about technology, whether it’s “they are too big and squashing competition” or “they are going to take all our jobs.” What do you think explains this?
Well I think there are at least a couple of things going on. One is that periods of deep and fast change are inherently scary to a lot of people; we are not wired for change. The other one is, let’s not kid ourselves that everything is okay for everybody because of technology or anything else. You and I have both written and talked about how the middle-class in America really is under a lot of pressure. The workforce is polarizing. We have a great demand for very high-skill people, we have a great demand for entry-level people, but the classic good old-fashioned middle-class is absolutely getting hollowed out, and people have every right to be anxious about that.
I cannot see how we will ever have a large, stable, prosperous American middle class doing industrial era routine jobs, either physical jobs or cognitive jobs. I just don’t think that those days are ever coming back. That doesn’t mean that we’ll never again have a large, stable, prosperous middle class; it just means that we’ve got to think long and hard about what the new jobs are going to look like, how to bring them into existence, and how to support people with a different set of institutions and different kind of safety net in the world that we’re creating.
The combination of the yearning for the industrial era and coal mining jobs and/or thinking about smashing up the parts of the economy that are working best — those are two deeply bad ideas.
You’re colleague at MIT, Daron Acemoglu, writes a lot about technology and he suggests that you have two different kinds of innovation or technologies: One which replaces jobs and one which enables jobs and creates new tasks, new things for us to do. Do you think that we are getting the wrong kind of innovation right now?
I think there’s a lot of automation going on these days, and it’s really easy to understand why, if you have lower costs and higher quality from doing something in a more automated way, every rational business decision maker is going to choose automation. However, I think we do under-emphasize sometimes that technology is also creating opportunities for people at all different rungs on the skill ladder. So, you and I were talking earlier about this idea of a health coach. That might not be the highest IQ person out there, that might not be the person with the most elite education, again, that is an empathetic, compassionate person who can get another person to go along and comply with the treatment plan. That could be a pretty decent middle-class job.
So one of the things that Erik and I and our colleagues at MIT have done is set up what we are calling an “inclusive innovation challenge” to kind of highlight these cases where innovators and entrepreneurs are coming up with companies and organizations that provide economic opportunity for entry-level workers, for classic middle-class kinds of workers, because the narrative right now, with some justification, is all about robots taking jobs. And that is an incomplete narrative.
I always like to go to Twitter and ask people for questions for the guest and I got one from economist Adam Ozimek, who tweeted “Is this the back half of the chessboard yet? It doesn’t feel like the back half of the chess board.” He’s referring to a kind of a complicated metaphor, so I’ll just say what he’s asking is, are we entering the period of exponential growth? You and Erik seem to be saying that we are in time of accelerating change, but in many ways it doesn’t feel like it.
The second half of the chessboard, is to say it a little too simply, when things get really weird and crazy. And when I look at, not economic growth rates, but the kinds of innovations that are coming out of the technology sector, things are getting crazy. And so, the example of the computerized Go champion — we were not anticipating this. So, to answer that really good question, my intuition is that we are in the early squares of the second half of the chessboard. We are in the period when things are getting a little crazy and we ain’t seen nothing yet.
Will that progress ever show up in traditional GDP and productivity numbers?
It’s a great question because it should be showing up in those measures. As flawed as they are, it should be showing up. I am bullish on productivity growth over the next 5 to 10 years, even with the problems that we all know about with measuring productivity growth. So, I need to get with Bob Gordon and place a bet with him about labor productivity over the next decade. I think it’s going to rebound.
So a policy idea here is that if you’re worried that we’re not using enough robots, and we’re not productive enough, then maybe we should raise the minimum wage to encourage companies to invest in more machines. You know, whether that’s more kiosks at McDonald’s or what have you. What do you think about that idea?
Well, a lot of municipalities are running exactly that experiment. So, Seattle, Los Angeles, San Francisco, I think New York as well are ratcheting up the minimum wage really really quickly. So, we are going to learn what happens when they do that. What I am observing is that the research so far says that what those big fast raises in the minimum wage do, is essentially, lower total employment hours for the lowest paid workers. In other words, they are kind of doing the opposite of what they are intended to do. Now, there’s a huge controversy about this; other people have looked at the evidence and come to a different conclusion, and my read of the evidence is that big fast raises in the minimum wage are hurting entry-level workers. Now if that’s true, I don’t want to do that, because our problem, I believe, is not that we don’t have enough robots, it’s that we don’t have good enough jobs for entry-level, lower-level people.
What would you tell policymakers right now, who I think are beginning to wake up to the issues of technology and automation and sustained productivity growth? What should government be doing?
I tell them not to try to solve the problems of 20 or 40 years from now, today. So, people are a little bit too worried about the jobless economy, or when the robots take over everything, what do we do with human beings. You know, maybe you and I will live to see that day, but that’s not the day we are in.
We are in an era where the American economy still has jobs without exception month by month for more than 80 months. We are not anywhere in sight of peak labor in the American economy. The problem is that instead of kicking out the great American middle-class jobs, it’s kicking out lower-middle-class jobs, more precarious ones.
Okay, if you go grab the Econ-101 textbook off the shelves, you’ll get a series of great ideas about what to do about that situation. And they include things like revamping our educational system, investing really heavily in infrastructure, trying to increase rates of entrepreneurship, being welcoming to immigrants as opposed to hostile to them, and investing in good old-fashioned basic research which turns into companies and jobs somewhere down the road. Any economist, if you woke them out of a deep sleep, could rattle those things off. My frustration is that for a while now, we’ve been doing a moderate to lousy job in all of those areas, and I think we are heading in the wrong direction more quickly now than we were before.
There are no comments available.
1789 Massachusetts Avenue, NW, Washington, DC 20036
© 2018 American Enterprise Institute