This expert thinks robots aren't going to destroy many jobs. And that's a problem.

By: Timothy B. Lee

Javier Zarracina

If you follow news out of Silicon Valley, it’s easy to get the impression that we’re living through an era of unprecedented technological change. The modern smartphone is less than a decade old, yet it has already become ubiquitous in American society and allowed Apple and Google to become two of the most valuable companies in the world. And companies like Uber and Airbnb are bringing the internet revolution to industries as diverse as taxicabs and hotels.

Yet it’s hard to find evidence of rapid progress in the nation’s economic statistics. To the contrary, after the productivity of American workers grew by an impressive 2.9 percent per year between 1995 and 2005, it grew by just 1.3 percent between 2005 and 2015.

That’s far slower than the productivity of American workers grew, on average, during the 20th century. And slow rate of progress is a major reason that middle-class wages have been stagnant over the last decade.

nme chart tim

So has the weak economic performance of the last decade been an unfortunate anomaly? Or should we get used to a slower pace of economic progress?

In an important new book, economist Robert Gordon makes the case for pessimism. He believes that technologies like smartphones, robots, and artificial intelligence aren’t going to have the kind of big impact on the economy that earlier inventions — like the internal combustion engine and electricity — did.

While new technologies like smartphones and Facebook have become a big part of our personal lives, Gordon writes, they’re not transforming the way we work. For all the hype about robots taking peoples’ jobs, there’s been hardly any progress automating labor-intensive industries like retail, medicine, or construction over the past few decades.

That might seem like a positive development if you’re worried about losing your job to a robot. But it’s a cause for worry if you’re hoping for bigger paychecks over the next decade or two.

We spoke by phone in early July. The conversation has been edited for length and clarity.

Timothy B. Lee: A lot of people think the US is going through a period of unprecedented technological change. What do you think they’re missing?

Robert Gordon: In my view, the big impact on business productivity from the digital revolution occurred between 1980 and 2005. That truly did make a difference as we got rid of typewriters and calculating machines and moved over to personal computers with spreadsheet software and word processing software.

Shifting away from paper forms that had to be sent by mail made an enormous difference in productivity. But that entire transition — and all of the resulting productivity growth — was basically complete by 2005. When you go to a doctor's office to check in, when you check out at the supermarket, you're interfacing with equipment and programs that are at least 10 years old.

Of course gradually there's an upcreep in the percentage of buying that’s done through e-commerce. E-commerce is 8 percent of total retail sales, but that still leaves 92 percent done in the traditional way. It's not that we don't have lots of innovation, it's that the impact on productivity in the business economy is limited.


TBL: E-commerce seems like a case where there is a lot of potential for progress due to self-driving cars.

In your book, you acknowledge that self-driving cars will have some direct benefits in terms of more convenience for consumers and that they could replace some jobs for truck drivers. But you’re skeptical that they will have larger economic effects.

It seems to me that if Amazon had a fleet of self-driving delivery cars, it could allow them to reduce delivery costs, offer delivery times measured in minutes rather than days, and significantly expand their share of the retail market. That could have a big impact, right?

RG: Almost all truck drivers have two roles. They drive the truck to the destination, they get out, they take the goods out of the back, load them on a pallet, and wheel it into the shelves. The people arranging the beer and the cola on the shelves are often truck drivers from the wholesaler.

Having a self-driving truck is not going to eliminate that role. You might have to have employment reorganized so that there are additional employees in the stores to take the merchandise off the trucks.

But basically my comment on Amazon self-driving cars is okay, the car drives up in front of my house. How does the package get from the Amazon car to my front porch? Who carries it up when I’m away from home?

TBL: But the "truck" doesn’t necessarily have to be truck-sized, right? For example, you could have self-driving vehicles that are small enough to carry just one package — maybe it just waits on your sidewalk until you get home. Or maybe there’s an Amazon drone that carries it from a nearby distribution center and drops it on your back porch. It seems like a mistake to assume an autonomous delivery vehicle will look exactly like a FedEx truck.

RG: Consumer Reports did a very detailed analysis of self-driving cars and pointed out that they're not ready for primetime. Self-driving cars can't see lane markings when there's snow and rain. We saw the tragic death of a tech enthusiast who was blindsided by a truck and the car didn't see it. The entire world has to be mapped in 3D, not just two dimensions, for cars to recognize all of the obstacles, all the signals, all the lights, all the construction zones.

The self-driving car may eventually take over, but we've got hundreds of millions of existing cars, all of which are going to have to co-exist with these self-driving cars. You can't make people buy them. And if self-driving cars are more expensive, as they surely will be because all the extra equipment that’s included with them, you'll still have a mix of autonomous and person-driven vehicles for quite a distance into the future.

In my book, I don't forecast anything past 25 years. I think the innovations that are going to be dominant for the next 25 have pretty much all been conceived even if they haven’t all been worked through. I think a complete transition to autonomous vehicles is much further off.

TBL: It seems like there’s actually two questions here. One is how long will it take for self-driving cars to become a mature technology that’s widely deployed. The other is how big of an impact will they have on the economy once that happens.

Is your view that the ultimate impact of self-driving cars will be small, or do you think it could be big but it’s going to take more than 25 years to get there?

RG: I think there will be gradual impacts. I think the biggest impact on productivity of self-driving vehicles will be long-distance truck drivers, where the job of driving the car on interstate highways will be taken over by self-driving features of vehicles that then may arrive at a staging area and be driven into the final unloading process by humans. That would save an enormous amount of time.

I think self-driving taxis are further off because of the variety of locations that taxis are expected to go that self-driving cars won’t know how to go to. The combination of Uber summoning with self-driving cars could work together to make an Uber-type autonomous vehicle come to fruition before a traditional taxi that has to look for someone waving at them on the streets of Manhattan.

TBL: It seems like one of the big factors driving lower productivity is the economy-wide shift from goods to services.

A lot of services have an inherent limit on how efficient they can get because the service is being provided by a human being who can only serve so many customers at a time. Do you think that’s a useful way of thinking about the long-term trend here?

RG: I do. I think manufacturing has been much easier to automate. I think there are limits in the ability of robots as they're currently built and conceived to do things like hotel room service and making beds and folding laundry, and doing many of the jobs in the service sector in education and the gigantic health care sector.

Even something as mundane as the employees in hospitals who push you down the corridors from your room to an x-ray machine or an MRI test, that's all still being done by humans. You'd think that would be the first job to be replaced by robots, but there’s nothing like that happening yet.

There are intermediate industries like wholesaling that seem to be the next easiest to automate. There are robots in warehouses bringing shelves to people at Amazon who pack your goods.

The Amazon warehouse is interesting because they don't go find the product, they go and find the shelf and they bring the shelf to the human packer rather than bringing the object. Presumably that’s because the robots are not yet able to pick up, physically, items that have such different sizes and shapes. Presumably that's a matter of time before that's improved.

But we've had industrial robots in manufacturing for more than 50 years. The first robot was introduced by General Motors by 1961. When I took tours of auto factories 20 years ago, robots were already ubiquitous. It's now 20 years later, and we've got even more robots in manufacturing.

So it's not that the robots aren't making this transition. It's just that they're not capable of doing many of the human tasks that are carried out every day in the service sector.

TBL: The economy grew surprisingly quickly between 1995 and 2005. Then over the last decade, it has grown surprisingly slowly. Now you’re forecasting that the economy will grow just 1.2 percent — well below the 20th century average — over the next 25 years.

I wonder if there’s a danger of overweighting recent experience in making projections like this. Isn’t it possible that there are forces that could start to speed the economy up again over the next decade?

RG: This idea that we're on the verge of some kind of fundamental transformation implies that there's going to be a major replacement of people by machines across the wide span of the service sector. And we just don't see it. We don't see the foundations of the innovations that would make this possible.

Look across retailing. Look across medical care and education. Look at finance, which had its big computer revolution back in the 1980s and 1990s — where everything is already being done by computers. We’re not seeing anything on the horizon that's analogous to the movement from typewriters to personal computers that was the big driver of productivity growth.

TBL: I’m currently reading Kevin Kelly’s book The Inevitable, and he paints a pretty optimistic (or pessimistic, depending on your perspective) picture of the future of on-the-job automation.

He predicts that fruit and vegetable pickers will be replaced by robots. He thinks pharmacists will be replaced by robots. Truck and taxi drivers will be replaced by self-driving cars.

Obviously you don’t think very much of this will happen — at least in the next 25 years. What do you think he’s getting wrong?

RG: I don't see any sign on the horizon that there are robots in the works that are going to count out pills in the back of Walgreens or CVS pharmacies. That’s just not on the horizon in the near term. That is measured in decades.

Robots are having trouble standing up without falling over. Robots are having trouble walking up and down stairs. Robots don’t have the dexterity to open bottles of pills.

You'd have to have a complete redesign of the whole pharmacy. You'd have to move away from the current technology that has shelves with big bottles of pills where they're counted out into little bottles.

That does seem awfully primitive, but it also seems primitive that supermarket shelves are stocked by humans. Where are the robots coming to replace humans stocking shelves in supermarkets?

Now, of course, we do have predictions that with the rise of the minimum wage, that we're going to have more installation of automated machinery in fast food restaurants. I fully expect that to happen. That would create an increase in productivity in the restaurant sector.

So I agree that digitalization and computers are going to gradually replace many jobs. But I think the word is gradual. There are big obstacles in the way of physical robots taking over many of the physical jobs that people do in their everyday lives.

This story is part of The new new economy, a series on what the 21st century holds for how we live, travel, and work.