Automation has displaced a lot of workers in the last 50 years, and it’s set to displace a lot more of them—taxicab and truck drivers, once vehicles drive themselves; much of what remains of manufacturing and assembly work and maybe even a lot of construction labor; fewer lawyers and doctors, once Watson-like software is perfected; teaching, except for the few people making the videos that everyone else learns from. Will we even need waitresses, or just people to bring out the food that we’ve ordered ourselves, once iPads replace menus?
The endgame here is the so-called singularity—the point at which technological development, spurred by Moore’s Law and another generation or two of software and robotics development, is so sophisticated that humans have become irrelevant.
An article in The Atlantic Monthly, back in October, by Rice University professor Moshe Vardi, tackled the question of whether that future is inevitable, and if so, what will it be like. He wrote: “[Artificial intelligence’s] inexorable progress over the past 50 years suggests that Herbert Simon was right when he wrote in 1956, ‘Machines will be capable...of doing any work a man can do.’”
Vardi continued, “I do not expect this to happen in the very near future, but I do believe that by 2045, machines will be able to do if not any work that humans can do, then a very significant fraction of the work that humans can do.”
Moshe Vardi is the Karen Ostrum George Distinguished Service Professor in Computational Engineering at Rice University and directs the Ken Kennedy Institute for Information Technology there. He’s an ACM Fellow, a AAAS Fellow, and most importantly, an IEEE Fellow, and he’s my guest today by phone.
Moshe, welcome to the podcast.
Moshe Vardi: Thank you. It’s a pleasure to be talking to you.
Steven Cherry: You cite a book that came out last year, Race Against the Machine, by Erik Brynjolfsson and Andrew McAfee. I interviewed McAfee last year on this show, and the examples I gave in my intro came mainly from that conversation. What do you think are some of the most vulnerable professions between now and 2045?
Moshe Vardi: Well, I think it’s interesting that you mentioned in your introduction waiting on tables, because just yesterday I saw an article with some videos from restaurants, where robots are now delivering the food. So this is already happening.
But I think what we are going to see happening are certain more-routinized jobs. Manufacturing is coming back to the United States, but without employees, practically, so the plants that we will see in the United States will mostly be automated plants. Even Foxconn, who is an employer probably of a million manufacturing workers in China, is investing heavily in robotics. So I think manufacturing will become automated to a larger and larger degree, not only in the developed world, but also in the developing world. And the price and the cost of these robots is going to go down.
Driving is a huge source of jobs. Just think of how many people drive, you know, transportation, goods, much of it we’re going to see automated in the next two decades. I believe that it’s going to be quaint in another generation, [to] talk about driving your own car. It’s just going to be quaint.
And, I mean, more and more, you know, you go to supermarkets, and more and more there’s automated checkouts. And then even loading the shelf will be automated. Warehouses, you know, maintaining warehouses will be automated. So as both AI and computing power and robotics all continue to progress, we will see bigger and bigger swaths of jobs just being taken over by robots.
Steven Cherry: Yeah, I would imagine that the driving thing is even going to affect farming. Between GPS and self-driving vehicles, there won’t be much need for people there.
Moshe Vardi: Yes, so even farming now is really not a major employer anymore. I think it’s just 1 percent of the economy. Used to be, I think if you go back 100 years, it used to be 85 percent of the economy. But whatever little, the fewer jobs that are left in farming, again, we will see them being gradually eaten away by automated machinery, with GPS, with self-driving machinery. Yes, much of it again will be automated.
Steven Cherry: Yeah, I was thinking more the third world in terms of the impact there. You mentioned in your Atlantic article an article that Wired published in 2000, by Bill Joy, in which he calls robotics, genetic engineering, and nanotech “our most powerful 21st-century technology.” It’s interesting that Bill Joy of all people doesn’t mention Watson-like software, or software at all.
Moshe Vardi: Yeah, it is interesting. And, you know, I actually remember when the article came out. Bill Joy is the ultimate techie, so for him to express that thought was almost heretic. You know, I think the article at the time was widely cited, but I don’t think it was widely influential. I think most people kind of read it, in my opinion, and shrugged. I don’t see that there was much of a follow-up on it, of people really questioning it.
But I think what happened in the last decade, especially in the area of robotics and software automation, is such huge progress has been made, and people start weathering the big recession, which made people very sensitive to the issue of jobs. And economists started paying attention to the fact that really, we’re now seeing a trend of about 30 years that includes the compression of income, for middle- and low-income workers, growing inequality in the economy, a greater and greater portion of the economy going to capital vs. labor. So this has become over the last year a very, very hot topic among economists to understand what’s happening to the job market. And so suddenly this issue that I think 10 years ago everybody was shrugging Bill Joy’s article off, suddenly it’s become something that’s on everybody’s mind.
Steven Cherry: So what professions will still be around in 2045, and in particular, what are the remaining sweet spots for our listeners, the sort of people who today are engineers and scientists?
Moshe Vardi: You know, it’s really hard to say, because there are things that computers do that we really never imagined them doing. You know, we can think of the things that really require soft skills. Okay, the more it requires human contact and soft skills, the more I think this is going to be the piece that we’re very far from having.
You know, we’re very far from having robots as salespeople that make cold calls to convince people to buy something. We are very far from this, okay, because these jobs really require human-to-human contact. And you can go on now, in my daily life, I look at jobs and I say, okay, in this job how routinized is this job? What level of proprietary skills and motor skills does it require? And you see as you look at jobs, you say, well, yeah, I think in 10 years this can be automated. I mean, just think of the people who work at the, as you go about your life, you know, well, tollbooth collectors, automated. You know, checkout clerks in shops, automated. One job after another will fall down.
Steven Cherry: In a short talk you gave in England, you asked recently what does humanity look like in a world where machines can do things better than we can? So what does humanity look like in a world where machines can do things better than we can?
Moshe Vardi: So I think, you know, we almost have to go to science fiction to try to figure it out. And there’s another prospect also, and in fact the counterpoint to my worry is Rodney Brooks’s worry that we won’t have enough robots, because we’re going to have a gray population, and we’ll have many more older people than younger people, and we’ll need all these robots to take care of the younger people.
And I have to say, I look at this, and to me this whole vision of a graying society where people don’t even feel that they necessarily have to have children because they will be taken care of in old age by robots, it’s not a great prospect for humanity. So, and this is something that we need to, you know, these can’t be issues that people in our discipline, in computer science and engineering, we don’t usually worry about these issues. We say, oh, these are matters for sociologists, okay?
But we seem to be blindly developing the technology without worrying about the consequences. The same way that there’s a lot of discussion, not much action but a lot of discussion, about global warming, even though when we’re talking about when things are going to become really deep, they’re going to be very adverse, it’s going to be, like, 2100, which is almost 90 years away. I think the impact of AI and robotics on jobs is not going to wait 100 years. It’s going to happen in the next 30 years.
Steven Cherry: My producer, probably not contemptuously, wrote me a note last week that said, “I don’t think all of them will pursue a lifelong interest in Kierkegaard.” I’m not sure what she has against Kierkegaard, but anyway, in that talk in England you concluded, “We are facing the prospect of being completely outcompeted by our own creations.” Now, I think Kierkegaard would have had a lot to say about that, but 60 years ago this future didn’t seem inevitable. Is it inevitable? Is there any way of getting off the robotics and artificial intelligence freight train before we reach singularity station?
Moshe Vardi: So in this society, we have not done, I think…we have not been able to, I would say, generally, to regulate technology. I mean, the technology seems to, I mean, we adopt technology, we discover the consequences later, and at that point it’s very often too late to get off the technology.
I mean, if we look at automobiles. You know, automobiles brought us an immense amount of convenience and flexibility. And now we are thinking that these automobiles are, what they have done to us, what they are doing to climate, what they are doing to air, what they have done to our cities, can we get off automobiles?
It’s just particularly impossible to do it now. But generally we run away with technology. I mean, the only community that I can think of that have a very conservative attitude about technology are the Amish, who consider the consequences felt before they are willing to adopt it. But, you know, most of the rest of us think of them as quaint, and that’s not the prevailing attitude in society. We run away with technology and deal with the consequences later.
Steven Cherry: I’m glad you framed the issue that way, because in conclusion I kind of want to ask you about your own take on this. I don’t know how old your children are, if you have children, but if you had the choice, would you prefer your child to be born in 1984, 2004, or 2024, and I picked those years because I want to know, basically, if you would prefer your child to turn 21 in 2005, 2025, or 2045.
Moshe Vardi: So I have a child who’s in his mid-30s, and I had some time ago a talk asking, basically, are we, humanity, happier today than we were, let’s say, 100 years ago, 200 years ago? It was by Stephen Jay Gould. And he said that he had a hard time convincing himself that we are really happier, that we are really better off because of technology. Except for one aspect: He said he cannot imagine we are not better off, and this is child mortality.
So he said you went back 100 years ago, and you expected several of your children to die before reaching adulthood. And today when a child dies, it’s an unexpected tragedy. And he said he could not imagine that this was not hard even 100 years ago, even if this was the norm. So I’m very happy that I have only one child, and I’m very happy that the child mortality is not a serious, prevailing risk these days. At least the world now is familiar to me. And I think it’s under control. The world in 50 years is, either will be a utopia or a dystopia.
Steven Cherry: And which will it be?
Moshe Vardi: Ah, that we don’t know. This is the big mystery of life. We don’t know how the world is going to unfold. And we have to contend with both possibilities. I think what we could do today is start having conversations about the future.
So we are, for example, with climate change, there is a vigorous conversation going on, okay? I mean, there are people who accept it, there are people who deny it. We are discussing countermeasures, we are discussing consequences. We haven’t done much action, but at least it’s a topic for discussion.
I think the issue of machine intelligence and jobs deserves some serious discussion. I don’t know that we will reach a definite conclusion, and it’s not clear how easy it will be to agree on desired actions, but I think the topic is important enough that it deserves discussion. And right now I would say it’s mostly being discussed by economists, by labor economists. It has to also be discussed by the people that produce the technology, because one of the questions we could ask is, you know, there is a concept that, for example, that people have started talking about, which is that we are using, we are creating technology that has no friction, okay? Creating many things that are just too easy to do.
I saw an article just last week, I think about electronic dating. All these dating websites, and it says, “It’s too easy, not enough friction.” And so it’s just too easy to find people; it’s too easy to dump people. It’s not like real dating, and so maybe we need to start thinking, as we’re producing technology, about using human technology, and to think about the consequence of technology, and maybe who has a role in producing technologies that will mitigate its impact. Right now we’re not doing it. We’re producing technology, damn the impact.
Steven Cherry: Very good. In the words of another great 20th-century existentialist, Yogi Berra—I guess this is this show’s favorite quote; I think we’ve used it three times now—“Prediction is hard, especially about the future.”
Moshe Vardi: And when you come to a fork in the road, take it.
Steven Cherry: Exactly. Well, thanks for taking a shot at prediction, and thanks for joining us today.
Moshe Vardi: It’s my pleasure.
Steven Cherry: Very good. We’ve been speaking with Rice University [computational] engineering professor and IEEE Fellow Moshe Vardi about the future of work and leisure, and war and peace, in the year 2045.
For IEEE Spectrum’s “Techwise Conversations,” I’m Steven Cherry.
NOTE: Transcripts are created for the convenience of our readers and listeners and may not perfectly match their associated interviews and narratives. The authoritative record of IEEE Spectrum’s audio programming is the audio version.