“I’ve got some bad news for everybody,” Missy Cummings says. In 2038, “you might be able to get groceries delivered to you by a slow speed robot — but you are
still not going to be able to call a car on your phone, jump in the backseat, and have it take you to Las Vegas.” Cummings, the director of the Duke University Humans and Autonomy Laboratory, is not “anti-hype” — just realistic. “Driverless cars are still kind of the Wild Wild West of development,” she explains.
On the latest episode of 2038, Intelligencer’s podcast about the future, Cummings talked with Max Read and David Wallace-Wells about what we can (and can’t) expect from robot-driven vehicles over the next 20 years.
So you’re saying no flying cars.
Well, you know, flying cars are tricky because they are possible. And certainly we have all the technology in place to do that in 2038. You might be able to get in a flying air taxi, potentially — one that is also a car in another country. And you will have to have a lot of money. So maybe in China or Dubai you’ll be able to take these services, but they will not be for the common man.
What are the big problems with driverless cars?
So, driverless cars are still very, very immature technologies. The fundamental automated technologies that power drones and rail, for example, are quite mature. We’ve had for a long time, and we actually know how to implement them, and how to implement them safely. Driverless cars are still kind of the Wild Wild West of development, and researchers are still learning new things about how they reason, how to make sure that they’re not so brittle — which means that they break under very unexpected and often very benign circumstances. I mean, this is why Tesla autopilot is so dangerous, even though it’s a deployed technology: Today, millimeter wave radar at highway speeds cannot detect static obstacles. This is why we’ve seen some deaths in Teslas, because the radar literally can’t see what is very obvious to the human eye.
When you hear from super optimists on driverless cars the idea is that they’re going to be flooding America’s highways in the next decade. Is that because those people don’t see the problems with the algorithms and the technology that you do? Or is it because they just think that computing power is going to get so much better over that period?
I think the answer is both of those things. I think there’s a lot of misunderstanding. I think there’s a ridiculous amount of technological illiteracy running amok in what would presumably be otherwise smart people — even the people who are running these companies themselves. I do a lot of c-suite consulting. And for the most part I have made quite a lot of money and a reputation out of running around and telling CEOs they do not know what they’re talking about, and they need to quit saying things like “artificial intelligence” and “blockchain.” So, yeah, I think the main thing is that there’s just a gross amount of misunderstanding. I’m not totally anti-hype — hype is good. This is actually where I think the technological illiteracy is a real problem. You do need some hype to keep excitement and motivation, and keep your stock price moving along. But what worries me is when people can’t differentiate. Okay, we’re going to put some hype out there just to keep investors interested and excited, and we actually know that it’s hype. But I think what’s the problem is happening now is that we’ve got CEOs — very well-known, well-spoken people — who are putting this stuff out there and they actually believe it. And it’s just simply not true.
I’m interested in why you think we’re going to see automated cargo plane flights but not automated commercial passenger flights.
Oh, heck no. Because you always have drunk passengers, right? Wherever you have groups of people, you need the James T. Kirk to manage those people. So you’ve got to have somebody in charge, whether it’s the captain as we know it today, or maybe it’s the uber-super-stewardess or flight attendant of the future, who’s both the captain and a trained U.S. air marshal. And also can get you drinks. When you think life is on the line, there’s actually a name to that. It’s called Shared Fate. Most people would not fly on a passenger or drone aircraft because they want somebody in the front who shares their own fate.
Is traffic going to get better by 2038? In L.A., or anywhere?
Look, if we all had our driverless cars, what we would do is have our driverless cars take us to work, drop us off, and then go drive around in circles until it’s ready to go back again. And then the problem we’d create would be so much worse, right? And, you know, urban planners want to say, Oh great that will free up all kinds of parking. Does that mean my car has to go all the way home and sit in the garage? Right. I mean, so, how good is that? Now I’ve doubled you know my carbon footprint for my commute to work. So I can see a real dystopian future where we think all these cool technologies are going to amazingly improve our life. And then what we create a monster behind-the-scenes. That’s one extreme. I do think that these traffic jam autopilots — if we could get those moving you would be amazed at how much better traffic would flow. I would guess, given some of the simulations that we’ve done over the years, that if traffic jam pilots became standard on all cars just that act alone would keep traffic congestion from getting as bad as it is right now, so you would see a dramatic improvement. But I think we need to be careful when we start to roll out these technologies that we think through all the ramifications.
To hear more, click above to listen to the episode in full. You can subscribe for free wherever you get podcasts.