Article Attribution and Read More at: Article Source
Smart machines — call them drones, autonomous vehicles, algorithms, or robots — have the potential to dramatically alter job markets everywhere. Or, rather, keep altering them since the these changes are already happening.
I’ve written about this repeatedly (see posts here, here, here, and here.) But for more answers, we turn to Martin Ford, software developer, entrepreneur, and Silicon Valley-based author of The New York Times Bestselling Rise of the Robots: Technology and the Threat of a Jobless Future and The Lights in the Tunnel: Automation, Accelerating Technology and the Economy of the Future. He has a degree in a computer engineering from the University of Michigan, Ann Arbor and a graduate business degree from the UCLA.
Here are some edited excerpts from our conversation via my podcast at Ricochet:
James Pethokoukis: There’s a lot of economic reports saying that US productivity and innovation isn’t what it used to be. Can we worry about a technological slowdown at the same time we’re worried about robots becoming smarter and more capable and taking all of the jobs?
Martin Ford: Well, I think we can. I mean, the interesting thing that’s happened, I think, over the last 30 years or so is that all of the progress has really focused in the information technology arena. I’m sure anyone who’s around my age remembers, the days when we anticipated advances in the space program and we thought there would be all this amazing stuff in the future, flying cars and all of that. And of course, most of that stuff has really turned out to be kind of a disappointment, except in information technology. If you think back to “Star Trek,” we don’t have any of that stuff, but we do have the communicator and we’re getting to the point where we may have the computer.
So what that implies, I think, and I think part of our problem is that all of that information technology is really having an impact on job markets. And one thing this is doing is it is making things less labor intensive, and that’s beginning to show up in some of the statistics, for example, in stagnant wages and so forth.
Is [technology] accelerating? How much should we be concerned about automation? Even if [there is acceleration] in information technology, shouldn’t that be reflected in statistics like productivity? If machines can do more, whether literally talking about robots or talking about software, if machines can do more and they are replacing people, shouldn’t we also be getting more productive that way? How come we don’t see any of this advanced robotics in productivity statistics? Or are we just missing something?
Well, I – you know, that’s something of a paradox and I think a lot of economists have been asking that question. There are a couple of theories out there. One, of course, is that there is a lag between the time that the technology comes online and whether businesses are able to really assimilate that and whether it shows up in the statistics. And of course, we did see rising productivity earlier in this century and now it has sort of petered out. So perhaps it kind of goes in bursts, perhaps.
You know, right now, you’re seeing some really remarkable advances on the technical side, in areas like artificial intelligence and what’s called deep learning or machine learning. And the question is, is that going to show up in the productivity? And I suspect that it will. Perhaps there will be something of a lag.
Another issue that I talk about that the economists don’t focus on too much is that, you know, productivity is tied to output. I mean, productivity, essentially, is the value of what we produce divided by the number of hours to produce it. And people tend to, when they talk about the robots, they focus just on the hours worked. In other words, they focus on the denominator of the productivity expression. And they assume that once the robots come online, the number of hours worked is going to fall and productivity should sore.
But you know, there is also an impact on the output, on demand. If there are fewer people out there that really have the income and the purchasing power to drive the economy and to create sufficient demand, then we may actually produce less than we might optimally. And that also could influence productivity.
So there is clearly a feedback loop there. If you automate some of the jobs, you can’t assume that productivity is just going to soar because that doesn’t happen in isolation. We only produce things in response to market demand. So if automation actually kills off some of that demand, then, the impact may be a little less clear than what most people expect.
Just give me a few examples of the sorts of advances that you find sort of most impressive and where those might lead us as far as either machines replacing humans or maybe in some ways complementing what we do.
Well, I think one obvious example is self-driving cars. I mean, when I started writing about this, back in 2009, I would have said that that was pretty much science fiction. That was something that in the longer-term future we might see. But the idea that a car could drive itself in traffic seemed to be well beyond where we were in terms of technology at that point. But obviously, Google has come pretty close to solving that problem. And now every other manufacturer has basically jumped on that bandwagon and does look like we’re doing to see progress over the next 10, 15, 20 years. I think it’s inevitable that we are going to have cars that can drive themselves.
And if you look at the approach that Google took to solving that problem, which is basically to use huge amounts of data and to map it out in advance, I think that that general approach can be applied all over the place. It’s all about machine learning and data. And there are probably many kinds of professions that are actually easier to automate than driving a car. If you are sitting at a desk doing the same kind of routine work again and again, if you’re a knowledge worker, you produce formulaic reports or some type of analysis, all of that is ultimately going to be susceptible to this.
So you can see on a very, I think, broad based impact from this. I mean, it’s really going to be across the board. And the thing is that it’s going to impact jobs really at all skill levels. It’s going to be blue-collar jobs driving cars or maybe working in fast food flipping hamburgers. Those jobs are certainly going to be impacted. But white-collar jobs taken by college graduates are also going to be impacted. So I think it’s a very broad based phenomenon.
I think in more than half the states, like nearly 30 out of 50, the single most common job is actually truck driver, either big rigs, delivery trucks. And that’s startling because that really does seem to be a job that could be greatly automated.
That’s right. And more broadly than that, it turns out that about 90% of our workforce works in occupations that existed 100 years ago. Now – and of course, it is. It’s areas like driving vehicles. It’s working in food service. It’s working in offices doing relatively routine things. Maybe working in factories, working in warehouses. These are all jobs that were there 100 years ago.
And so you hear a lot about all these new jobs being created by technology, like the website designers and the social media marketers and so forth. But those are a tiny number of jobs. Most people are working in traditional areas and a lot of those areas are going to be susceptible to this. And that’s just millions and millions of jobs that could essentially be vaporized.
The service sector, again, seems a little more difficult since it’s a lot more human interaction. But you obviously think the service sector’s obviously also right for this kind of disruption.
I think that’s the big disruption that’s coming. I mean, what you say is true. The service jobs traditionally have been harder to automate and that’s why we now have a sector economy, right? Everyone basically works in the service sector. Manufacturing is now less than 10% of employment in the United States, whereas back in 1950 it was more than a third.
But I do think that the service sector is next. It’s what’s coming next. You know, one of the things that’s really disruptive is this technology we call machine learning, which is all about analyzing data and having the machines or the algorithms figure out from this for themselves how to do these jobs. As opposed to what you just said, which is having a programmer figure out, you know, what are the 12 steps to do this job and then trying to sit down and program that.
So we now have this approach where the machines are figuring it out for themselves, and that’s going to be especially applicable across the service sector. But, you know, there’s lots of evidence that this is going to happen over the next 10-20 years and there’s going to be just an enormous impact on those service jobs.
The second chapter of your book is called, “Is This Time Different.” I’m sure in every interview you get the traditional economist’s analysis, that, yeah, we’ve heard this story before. And even though technologies can be disruptive and over the short term there might be a lot of losers, over the longer term, we figure it out. We figure out how to create new jobs that machines can’t do. We train humans to do these new jobs. So we sort of stay, you know, a few steps ahead so we don’t have mass unemployment. And you don’t have to go back to Industrial Revolution or the Luddites.
But in your book, you [discuss] the automation fears that we saw in the 1960s; there were a lot of concerns that technology was advancing so quickly that, you know, we would have a jobless future. Now here we are, decades later, tens of millions of jobs were created, pretty low unemployment. So why is it this time different?
Well it does have a long history and it’s reasonable for people to be skeptical. I always say that there seems to be something in common between this and that old story of the little boy who cries wolf. The false alarm gets raised again and again and people become complacent and skeptical, but in the end of that story the wolf does show up, and I think that that could happen this time. And the reason is really two things.
One is that machines are now in a limited sense beginning to think. They’re beginning to take on cognitive roles. And rather than just moving into some new area of the economy or having some new specific capabilities, machines are now beginning to encroach on that fundamental capability that really sets human beings apart. The thing that so far has allowed people to stay ahead of the march of technology is our ability to think and solve problems. As machines increasingly move into that intellectual work, it’s going to be more and more difficult for people to stay ahead.
The second thing is that, you know, all of this technology, including information technology generally – and I think artificial intelligence specifically – is really going to become a general-purpose technology. It’s just everywhere. It’s ubiquitous. It’s not like the innovations we’ve seen in the past. Specifically, for example, you look at agriculture. Agriculture is the classic example. In the United States, most people at one point worked on farms. Now less than 2% of the people worked on farms, and yet obviously that wasn’t a bad thing. Millions of jobs were vaporized, but it didn’t turn out to be a bad thing at all. People did move on to other roles that were more fulfilling, that paid more. You know, food is a lot cheaper now than it once was as a result of all that. So it was a great thing.
And so the skeptic will say, isn’t this just going to happen again? Aren’t we going to see what happened in agriculture just unfold? It’s going to be the same process.
That’s a pretty powerful counterargument.
Right, and my point is that that was a specific technology. That was mechanical technology specific to the agricultural sector. And at the time that transition happened, there was the whole rest of the economy out there to absorb those workers, right? You had a rising manufacturing sector. So people moved from the fields into factories. And then later on they moved from factories to the service sector, which is where they work now.
But the point is that the agricultural technology did not invade the factories and the service sector. But today, we’ve got this ubiquitous, across-the-board, general-purpose technology that’s coming everywhere. It’s going to scale across the entire economy. It’s going to come for every employment sector. There really isn’t going to be a safe haven. There isn’t going to be – it’s very hard to imagine some new area, some new industry that’s going to rise up in the future and it’s going to be very labor intensive.
You can certainly think of new industries that will arise, nanotechnology, synthetic biology, all of that. You could think of plenty examples. But I would challenge anyone to think of an example of a new industry that’s likely to arise in the future that will need to employ millions and millions of average workers. The fact is that that story is probably pretty much over.
The other thing that I focus on is really the nature of the work that’s being done. I mean, you know, people tend to think of this in terms of industries. They think, well, this industry gets automated or disrupted by technology, but there’ll always be some new industry or employment sector that will arise. Yeah, but the point here is that most people come to work and they do things that are on some level fundamentally predictable. They do things that are, in a sense, routine. You know, not rote, repetitive, but people do the same kinds of things again and again, and the things that they do are predictable based on what they’ve done in the past.
And if that is the case, then it means that what they’re doing is ultimately going to be susceptible to an algorithm that can churn through data and learn from that and figure out how to do a lot of that stuff. That’s the essence of machine learning and that’s really, I think, the big disruption that’s going to come.
And again, the important thing is that it’s across the board. It’s not specific to one industry or one area. It’s just a general-purpose technique that’s going to be applied everywhere. And that’s really what’s different from what happened in agriculture.
I wonder if you could just talk a little bit about that, the bit of that chapter about the ’60s panic and what it has to do with Martin Luther King Jr?
Obviously, we have certain perceptions about Dr. King that he’s, you know, exclusively all about civil rights and so forth. But it turns out that in 1964, there was a formal report put together by a very smart group of people. It included two Nobel laureates. It was called the Triple Revolution Report. And it talked about three revolutions. One was the civil rights revolution, the second one was the revolution in nuclear weapons; and the third thing, which everyone has now forgotten about, was the revolution in automation.
And this report actually focused mostly on that third issue and predicted that industrial automation was just going to totally upend the American economy, that there was going to be massive unemployment and social upheaval and so forth. And this report was put together and given to President Johnson in 1964. And at the time, it was a big issue. People were really worried about it. And it turned out that a few years later, you know, Dr. King gave what turned out to be his final Sunday sermon in Washington National Cathedral and he actually talked about this.
He didn’t just talk about civil rights. He talked about the impact of automation on the job market. So this was an issue that I think a lot of prominent people, a lot of intellectuals at that time had kind of assimilated. They were thinking about it. And it’s interesting how it was such a big issue then, but now, you know, in the years since then, it just completely went off the radar. And of course, the reason is that it didn’t happen. I mean, it was predicted. It did not happen.
But again, my feeling is that these people essentially got the basic idea correct, but they were just too soon. They were worrying about this at a time when computers where these huge monstrosities that filled entire rooms and yet had, you know, dramatically less power than you now have in your cell phone. They were just these big, plodding machines that weren’t even close to having the kind of capability that you would need in order to actually bring this off and make this happen. But of course, people didn’t realize that.
And so generally, I think that’s been the mistake that has been made. I think we need to distinguish between two kinds of errors here. I mean, it’s fine to say, OK, these alarms have been raised and they’ve always been wrong. But there are two ways to be wrong. One is to simply be premature and the other way is to be fundamentally wrong, and in other words to – for the truth to be that this is something that can never happen. And I tend to not believe that.
I’m making essentially the same prediction. Maybe I’m wrong, too, but if I am wrong I still believe it will be in that same way. It’ll be that I’m pretty mature and maybe the technology isn’t there yet, and maybe this is 50 or 100 years out.
The other day, I wrote a blog post looking at the old-fashioned misery index, you know, from like the ’70s — at unemployment and inflation. You put them together and if that’s a really low number, people should be pretty happy. Well, that right now it’s a really low number. Unemployment’s way down, inflation’s low, but people still seem really unhappy and really concerned about the future. About two thirds of Americans think the country’s heading in the wrong direction. So I’m trying to figure out where that anxiety comes from.
How much of it is just the current economic situation and how much is it the fear that the future labor market, because of technology, is just not going to be very inviting for a lot of workers?
The fact [is] that productivity over the last 30 years has continued its relentless climb, although there have been periods of low productivity like now. If you look at it in the long term, I mean, it’s been climbing relentlessly.
On the other hand, wages for most average workers have completely stagnated. People are not getting a raise at all, I mean literally for decades. There are some groups of workers, people with lower skill levels now that actually make less in real terms than they did in the 1970s.
So a lot of people are correct in perceiving that they’re worse off or at least that they’re not making any progress…. this kind of pessimism about the future, and this idea that our kids are going to be, you know, worse off than we are and so forth.
And I do think that technology has a lot to do with that. I think it’s one of the – certainly, one of the main forces that’s creating that wage stagnation and the decoupling of productivity and wages.
What has worked in the past is a free market economy that was able to create a lot of new interesting jobs that humans could do and we just were able to train people to take those new jobs. Why won’t that process — a vibrant economy and just educating people better — work now and for the next 50 to 100 years?
There’s a fundamental transition going on in the nature of technology. Historically machines have been tools. They have been things that workers have used that have increased the productivity of those workers and as a result have made the workers more valuable. If you’re a worker and you learned to use a better tool and that took makes you more productive, allows you to produce more, then you can demand a higher wave. And that’s exactly that happened during the so-called Golden Age after World War II.
But I do think that there’s now a transition happening, and machines are transitioning away from being tools and turning into workers. They’re becoming autonomous. And rather than complementing people and making people more valuable, in many cases they’re actually substituting for people. And that’s not true of everyone, of course. There’s always going to be an elite group of workers who have the education and the skill level and the creativity and the talent to leverage the new technology. And those people are going to do fine. But my concern is that that group of people as a percentage of the whole workforce is likely to shrink over time… more and more people are going to be left behind.
And again, it gets back to the fact that most people are doing relatively routine things, routine predictable things. I mean, not everyone can be a rocket scientist and do – you know, you hear a lot about – you hear the word “creativity” come up a lot. The idea is that we need to retrain workers, so that they can do more creative, blue-sky thinking kind of things. And of course, that’s going to work for some people, but in the United States we’ve got a workforce of around 150 million people. I don’t think that approach is going to work for 150 million people. Not – you know, people have different talents, different capabilities. Not everyone can be trained to do that, you know, that really high-level intellectual type work and so forth.
So I do think that this is ultimately going to have an impact on a very large percentage of our workforce and we’re going to have to figure out a solution to that.
There’s a bit in the book where – President [Kennedy] was asked about this issue and what we need to do in the future. And he mentioned things like more job training, better education. And you wrote,
the President’s words capture the conventional and nearly universal assumptions about the nature of unemployment problems. More education or more vocational training is always the solution. With the proper training, workers will continuously climb the skills ladder, somehow staying ahead of the machine. They will do more creative work, more blue-sky thinking. There’s apparently no limit to what average people can be educated and trained to do. And likewise, no limit to the number of high-level jobs the economy can create to absorb all those newly trained workers. Education and retraining, it seems, are a solution that is immutable across time.
Can you pretty much train just about anybody to do very high-level, very intensive jobs: rocket scientists or creative, high science, high math kinds of jobs? How many people can really do those jobs?
Obama basically says the same thing. So the point is that, you know, our approach to this has not changed at all, but the technology has changed. I mean, back when Kennedy was talking, it was machines displacing manual jobs. Today, of course, it’s machine learning algorithms that are displacing intellectual work. So that’s the big change. And yet, you’ve got two presidents, you know, 50 years apart saying exactly the same thing in terms of what the solution’s going to be.
Again, the retraining thing is going to work for some people. I tend to believe that there are limits to human capability – that not everyone is equal intellectually or – I mean, maybe that’s controversial, but I don’t think that among academics that study this it’s controversial. It’s also more, as you know, of a conservative position to recognize that. If you talk to someone like Charles Murray, who, you know, has written about the so-called cognitive elite and all of that, that there clearly are differences in fundamental capability. I just don’t think that you can expect to train everyone to become an expert in machine learning or become even a computer programmer.
People have a range of talents and capabilities. And there clearly is, in terms of human capability, a normal distribution, you know? Most people are going to be clustered around average and a much smaller number of people are going to be exceptional. And the whole point I’m really making is that it’s not going to be good enough to have middle class jobs for the top 5% and everyone else is, you know, struggling to survive. That doesn’t make for a sound society.
And it also doesn’t make for a vibrant market economy. You know, we – if we get into a situation like that, we’re not going to have enough consumer demand out there. We’re not going to have enough people that can actually buy the things being produced.
So we need to figure out a solution that’s going to be much more broad based than that, where everyone is going to have access to opportunity or at least to a decent income.
McKinsey just came out with a study which looked at jobs by not broad categories, but what happens inside the job. And what they concluded is that fewer than 5% of occupations currently could be entirely automated using current technology. But about 60% of occupations could have a big chunk of what goes on inside that occupation, 30% or more of the activities automated. And so the conclusion was that the jobs aren’t going to be automated away, but that more rote, boring parts of the jobs would be automated away, and then people would do the more interesting jobs or bits of them.
Do you find that a persuasive counterargument that it’s a “racing with with the machine” scenario instead of against it?
I don’t completely dismiss it, but I find it confusing that people seem to perceive a huge difference between automating entire jobs or just automating tasks. I mean, essentially what they’re saying is it’s not – we’re not going to eliminate entire occupations, we’re just going to eliminate a huge percentage of the tasks done by those workers.
OK. Well, what does that mean? We can have a very optimistic take and say that employers are going to keep people around to spend half their time doing all of this really creative stuff. Maybe that’s going to happen in some cases. But the other scenario, of course, is that if you’ve got two workers and you eliminate half of what each of those are doing, then pretty soon, you – the employer’s going to figure out that they only need one person, right?
So I think that eliminating half of what each worker does obviously leads to a lower job count–
I think the counterargument would be is that those more creative tasks will expand because you’ll be focusing on them. You’ll be able to do more deeper thinking. You don’t necessarily lose the number of workers, just those workers will become more productive. But let’s say that the scenario you’re describing takes place. We have a small sliver of people – a much smaller sliver of people really doing sort of high wage, interesting work. What does that society look like, and what is sort of the policy response if education only gets you so far?
It would be an extraordinarily unequal society. I think it would undermine the basic hope that we all have. If education is no longer an effective way to guarantee a decent livelihood, then we’re really out of options. I mean, education is really the only tool we have in the toolbox in terms of working conventionally. So at that point we have a big problem socially, and I think we have a big problem economically because we’d have huge numbers of people that really don’t have sufficient income to drive the economy, to go out there and be vibrant consumers and keep economic growth churning.
So I – my take on this is that we would need a radical solution at that point. And what I’ve proposed or advocated is some form of a guaranteed income, you know, where everyone in our society will have access to at least some kind of a minimal income. I think that that’s probably the best policy response to this in the long run. And of course, I am talking about the long run. I mean, it’s not something that I anticipate happening today.
I think some of my Democratic friends are ready for it right now.
Well, I mean, it was advocated by people in the past. There are groups of people who think we should have a basic income now, and that’s not an entirely spurious argument I don’t think. But politically, it does seem impossible, certainly in the United States. It seems almost unimaginable. And the thing is, of course, is when you talk about a guaranteed income, the immediate reaction to that is that it’s socialism – that it’s a massive expansion of the welfare state and turning everyone into a taker and a slacker and so forth.
The interesting thing, of course, is that if you know something about it, you know that in the past it has been advocated by conservatives and libertarians, most notably, as I point out in the book, Friedrich Hayek. Because essentially what it is is it’s a free market approach to a safety net. Rather than having the government take over the economy and provide for people and build housing and feed people, or try to run industries in a way that artificially creates jobs and so forth, what you do is you give people an income and you let them go out and you participate in the market. So it’s actually a free market approach to providing a basic safety net. And that’s why people like Friedrich Hayek and also Charles Murray, for example –
My AEI colleague, Charles Murray has written a book making that argument, that it’s a more market oriented solution and in this sort of post-industrial economy, that eventually all countries will realize that’s the solution. Now, you’ve been on a tour for a while now promoting the book. What is the response? Do you sense the people are generally worried?
Yeah, in general, I do think people are very concerned about it. One thing I notice is that whenever I go and talk about this, if there is a question-and-answer period afterwards, usually I’m just overwhelmed, that we have to cut off the questions because there are too many. So people have a tremendous interest in this.
And I’ve talked to lots of different groups, including average, typical people who are really worried about. And I’ve talked to very technology-oriented people, including people that are actually working in areas like machine learning and artificial intelligence, and they’re worried about it, too. In general I found that among technical people, there is at least an emerging consensus that this is likely to be an issue as far as the job market is concerned.
There’s also this other issue out there about artificial intelligence that you hear from – recently, you’ve heard it from Stephen Hawking and from Elon Musk, this idea that there’s an existential threat that smart machines are going to come to life and take over and really threaten humanity. And that’s something that I think most people who are really working in the field don’t worry too much about in the near term. I mean, maybe 50 years from now or more that might be an issue. But for now, the issue that I think people tend to worry a bit more about is the economic impact and the job market impact, and the fact that more specialized technologies — technologies that are not science fiction — are going to have a very practical impact on a lot of the jobs that people now rely on for an income.
Maybe we should just genetically engineer people to be smarter. Is that a way to stay ahead of the robots? You know the Chinese are working somewhere to create smarter people.
That’s one proposal. And in fact, the Chinese are working on it. There is a research facility in Beijing that is specifically focused on trying to isolate the genes for intelligence. And I don’t think they’ve succeeded at that yet. I think most scientists who specialize in that would say that there’re lots of genes that impact intelligence, so it’s very hard to do that. But they are trying to do it. Obviously, that gets into a lot of areas that here in the West we would be very squeamish about. So it’s an area that in the United States and in most Western countries, I think that we’re going to be very reluctant to get into. Christian conservatives would be really very kind of put off by this whole idea, for example.
So I don’t really see that happening here, but you know, if the Chinese succeed at it, that would be a big deal. They could, over time, engineer a more intelligent society and they would have an enormous competitive advantage as a result of that.
Is there a particular science fiction movie that you think gets these issues right or is somewhat instructive on dealing with a more automated world?
I don’t know one that specifically focuses on automation and the job market. The dystopian science fiction movie that I guess made the biggest impression on me was “The Matrix” series. I think it really gets at virtual reality, which is another trend that I think is going to be important.
One thing – one question you can ask is if people really become unemployed and increasingly dissatisfied with the real world, what are they going to do? And I think one of the answers is they’re going to plug into the virtual world. And I think that that could ultimately become a real social issue for us.
I think if you’d look at some of the advances in VR, like the Oculus Rift, you’re seeing – this is in its infancy now. You know, what’s it going to be like 10-20 years from now once we apply the same kind of acceleration that we’ve seen in other areas? I mean, it could be, I think, something that’s just extraordinarily disruptive. It could be almost like a drug that –
But that’s the future you’re painting, a bunch of people sitting on their couch with their VR helmets getting their government basic income checks and there’s no work.
Yeah, I mean, that’s the dystopian output or the dystopian outcome, you’re right. And so it would be really important to design a guaranteed income in a way that we don’t have that kind of an outcome.
But aside from, you know, the issues of income and so forth, the technology itself of virtual reality, I think it’s just going to raise a lot of issues. I really do think that eventually, it may be something akin to a drug and may – you’ll have people calling for it to be regulated in a way that drugs are regulated perhaps. I think that we underestimate how disruptive it could be if people have that ability to enter a world that might in many ways be indistinguishable on some level from the real world and in that virtual world, they can have all kinds of things that in our world would require income and hard work and so forth. So it could actually just really kind of upend the way our whole society and economy works if it really gets to the point where it’s truly advanced.
Read More At: https://www.aei.org/publication/pethokoukis-podcast-will-the-robots-take-our-jobs-a-qa-with-martin-ford/