Baby, You Can Program My Car: A New Marketplace Podcast

Out for a ride with Jarrod Snider of Carnegie-Mellon's autonomous-driving lab: "The throttle, the brake, the steering wheel, the turn signals -- everything is being controlled by a computer right now. I'm just sitting here.”

Our latest Freakonomics Radio on Marketplace podcast is called “Baby, You Can Program My Car.” Yes, it’s about driverless vehicles. (You can download/subscribe at iTunes, get the RSS feed, listen via the media player above, or read the transcript here.)  

I recently had the good fortune to go for a ridealong in a self-driving Cadillac SRX4 with three of the engineers responsible for making it go: Raj Rajkumar, John Dolan, and Jarrod Snider, all key players in the General Motors-Carnegie Mellon Autonomous Driving Collaborative Research Lab. We rode around a large track that the university has built on the site of an abandoned steel plant in Pittsburgh.

The car's computers, housed where the spare tire would otherwise be stowed.

What was most remarkable, to me at least, was how unremarkable it felt to ride in a vehicle that no one was steering or braking. In other words, it felt normal — not like a science experiment or a rocket ride — and, as amazing a feat of engineering as a driverless car is, I also realized how much of the technology to go driverless already exists in the modern cars we’ve been driving for years (cameras, sensors, automation, etc.). 

That said, a lot will have to happen before we live in a world where people don’t do much of their own driving. But I can’t wait! In this podcast, we try to address some of the changes and gains that driverless cars will produce. (Surely there will be complications and downsides as well.) The Economist recently published an excellent summary of a driverless future, much more encyclopedic than this short podcast could cover. (See also a 19-year-old Romanian driverless inventor and the latest in Israeli driverless technology.) I guess we will have to come back for a longer episode.

To me, the largest potential gain by far is in lives not lost. While there have been heroic gains in auto safety over the decades, there are still about 34,000 traffic deaths a year in the U.S., and more than 1 million worldwide. In the U.S., traffic accidents send more than 2 million adults to the E.R. each year — and of course the economic impact is massive as well.

There are some things that computers will never do as well as humans. Driving is probably not one of them — especially since about 80 percent of drivers rate themselves above average.

The men who make the driverless car go (left to right): Raj Rajkumar, Jarrod Snider, and John Dolan.

Audio Transcript

Kai RYSSDAL: Time now for a little Freakonomics Radio. It’s that moment every couple of weeks we talk to Stephen Dubner, the co-author of the books and blog of the same. The hidden side of everything is what he does. Hey Dubner, how are you?.


Stephen J. DUBNER: Hey Kai, I’m great. Hey, I’ve got a question for you today: how well do you think that a computer would do you job, hosting a radio show? 


RYSSDAL: Ha! What are you kidding? You been probing my nightmares? What? That’s not a good question!


DUBNER: Here’s the thing, It strikes me we’re at this fascinating point in history where we’ve all become very reliant on computers, and yet there are some things that humans do – and will always do – better, like radio hosting. But let me ask you this parallel question: how good are you at driving, let’s say?


RYSSDAL: Well, I’m a man, so of course I say I’m a great driver. Right?


DUBNER: So, no matter how good you are, even good drivers obviously pose a risk. I was talking about this with a fellow named Raj Rajkumar. He’s an engineering professor at Carnegie-Mellon in Pittsburgh:


Raj RAJKUMAR Ninety-three percent of accidents happen due to human error. We really are trying to basically take the human error tendency out of the picture.”


RYSSDAL: Okay, 93 percent I get, but how do you “take the human error tendency out of the picture” man?


DUBNER: Here’s what you do Kai: you don’t let humans drive. OK, so Rajkumar heads up a Carnegie-Mellon team that’s been developing a driverless car. A couple weeks ago, I went for a ride this car.


RYSSDAL: Really?  I was about to ask if you got to drive it, but I guess not.


DUBNER: I didn’t get to drive, nobody got to drive it!


RYSSDAL: So did you go out on the city streets of, where’s Carnegie-Mellon... Pittsburgh, right?


DUBNER: We didn’t. It’s in Pittsburgh. In this case we just drove around a track that Carnegie-Mellon has built on the site of an old steel mill. But there were, on this track, there were other cars, bicyclists, there was a skateboarder in the road, there’s stop lights, there’s construction, all the elements of real driving.


RYSSDAL: What does this thing look like – does it have the big camera turret on top so it can see what’s going on, and sensors and all that?


DUBNER: Actually no. So that’s the Google car, that a lot of people have seen pictures of. So Carnegie Mellon is developing this car for General Motors, so this was a Cadillac SUV. And their mission is to build a driverless car that a) doesn’t look like a robot; and b) is relatively affordable. So all the cameras and sensors and radars are embedded in the bumpers and elsewhere. It looks pretty much like a stock car unless you open the spare-tire compartment, that’s where all the computers are. And then there’s also a big red “kill” button on the dashboard.


RYSSDAL: Aha. Is that for when the car’s about to crash into something and you’re sitting there going “Aaahhh!!”


DUBNER: Well, that’s the issue here. So GM and Google are not the only ones developing driverless cars. There’s a lot of competition – which I would argue is a very good thing – and from all the evidence so far, it appears to be astonishingly successful. At low speeds and high speeds, city streets, highways. And it looks like a driverless car will screw up a lot less than a car driven by us, by humans. So then the question gets to be: when do we get it? When does this happen for real, and what interests me really is what kind of effects will it have on society?


RYSSDAL: Well, spitball it for me. I mean, what’s going to happen?


DUBNER: Well, honestly, I personally think it’s a revolution waiting to happen. Just think about all the industries that get affected – for better or worse. The auto industry, of course, the insurance industry. Older people could live on their own longer if they don’t have to drive themselves. Drunk driving wouldn’t be such a big concern – which is good news for restaurants and bars. But to me the biggest impact by a long long shot is safety.


RYSSDAL: Yeah, because lots of people die in car accidents.


DUBNER: 34,000 traffic deaths a year, roughly, in the U.S. And if you look worldwide, 1 million deaths from traffic fatalities. Then there’s injuries. In the U.S., traffic accidents send more than 2 million adults to the E.R. each year -- and of course the economic cost of all this danger is massive. There’s also the fact that most people enjoy driving.


RYSSDAL: Pry this steering wheel from my cold, dead hands.


DUBNER: Well, that is a common sentiment. But the fact is that most of us don’t drive anywhere near as safely as we think. Get this Kai, about 80 percent of drivers rate themselves above average, which is, of course, statistically not possible. And believe me, if we found out that human error by, let’s say, public-radio hosts was causing 1 million deaths worldwide -- my friend Kai, I would replace you with a computer in a heartbeat.


RYSSDAL: Stephen Dubner, Freakonomics-dot-com is the website. We’ll see you in a couple of weeks.


DUBNER: Thanks for having me Kai


RYSSDAL: Alright man, bye bye.

Leave A Comment

Comments are moderated and generally will be posted if they are on-topic and not abusive.



View All Comments »
  1. eric smith says:

    regarding your comment on NPR about this statememt not being possible
    There are some things that computers will never do as well as humans. Driving is probably not one of them — especially since about 80 percent of drivers rate themselves above average.

    I think you mean above the median; it is possible for 80% to be above the mean. If you give a zero to someone who has an accident (say in 2013) and 1 to someone who does not then I think 80% will be above the average which should be between zero and 1 since there are probably fewer than 80% with accidents,

    Thumb up 6 Thumb down 2
    • Jon says:

      Yes, you’re right. Consider a hypothetical group of drivers who rate themselves from 1-10 based on driving ability… n=5

      [1, 9, 10, 10 ,10]
      Average = 8
      80% of drivers are above average.

      So this is possible if the bad drivers are really really bad. But I understand what is meant, everyone can’t be above average!

      Well-loved. Like or Dislike: Thumb up 5 Thumb down 0
    • TC says:

      I remember reading that Gutenberg put 70,000 scribes out of work overnight with the invention of the printing press. Taxi / Limo / Bus drivers will be a thing of the past when SDV is available. Personally, I look forward to the day when 700 to 800 people per week are not dying in the US from auto accidents. I also look forward to the day when my grandchildren will ask me – “So Grandpa, you mean you would drive the car all by yourself and you could go as fast as you wanted and even drive on the sidewalk if you wanted …..” I believe one very interesting outcome as the SDV takes hold will be when an insurance company rates you on the type of vehicle you use. They will provide very high premiums for conventional cars, but what about the driver operated car without insurance that hits a SDV? Or will any accidents with SVDs be a thing of the past too!

      Thumb up 0 Thumb down 0
    • Mike Melnyk says:

      Good catch. I was wondering if I was the only one who caught this math error… until now!

      Thumb up 0 Thumb down 0
    • Adam Smith says:

      I can’t understand why people think statisticians are smarmy know-it-alls who are more concerned with being precise than being correct. Where does that come from?

      Thumb up 1 Thumb down 0
  2. Russell Harris says:

    I had to read that last sentence twice – double negatives! It would have been easier to say that there ARE some things computers do better, and this is one of them.

    I read a quote from a US Navy pilot where she found that the auto-landing technology did a better job than she could, particularly on a pitching carrier deck. If we can do that, we can make driverless cars work.

    Another saving is from the insurance industry and car repairers.

    Plus also, I can imagine not having traffic lights – cars approaching an intersection would electronically “negotiate” who would have right-of-way based on safety, time and traffic congestion. You could even have people pay a premium for faster passage – paying money to those other people who would accept a slightly longer journey time.

    Imagine too a networked car would enable traffic planners to see exactly the start and end point of all journeys, and incorporate this into their planning. We are having this exact argument in my home town now about whether to spend $5B on a tunnel – no-one can say with certainty that this will help or hinder the problem in that part of the city.

    Well-loved. Like or Dislike: Thumb up 7 Thumb down 0
  3. Ephraim says:

    Your topic is timely, since I frequently see Google’s self-driving cars on the highways here in Silicon Valley. What I want to know is how the auto insurance industry will go about sabotaging what is so clearly a public good, since they can’t possibly like something that will cut deeply into their business.

    Thumb up 2 Thumb down 0
    • Rob says:

      I’m not sure that is necessarily the case. Insurance companies make money by properly quantifying risk and efficiently pricing against it. There should still be a need for that, even with far less risk.
      In the short term, the first insurance company who can accurately price policies for computer driven/assisted cars will make a huge pile of cash. They will be able to beat the price of competitors while minimizing their own risk. Insurers love to make underwriting profit but it’s hard to do while keeping a competitive price.
      In the long term, the prices of individual policies will fall and total revenues along with them but I still see a glimmer of opportunity. Driverless vehicles can also imply passengerless vehicles. We might actually end up with more, smaller vehicles on the road. They could be delivering groceries, dry cleaning, or pizzas to our curbs. With more vehicles and lower, more predictable risk, insurance companies will be able to stay profitable. The auto insurance companies might be better served by lobbying for a regulatory niche for themselves so that standard business owners policies don’t pick up all of that new market.

      Thumb up 3 Thumb down 0
  4. wholesale lingerie says:

    So this is possible if the bad drivers are really really bad. But I understand what is meant, everyone can’t be above average!

    Thumb up 0 Thumb down 0
  5. MW says:

    Speaking as one of the very rare below average drivers, I’m really looking forward to being able to have a self-driving car. It will be a win both for safety and stress on my part.

    Thumb up 3 Thumb down 0
  6. Jack says:

    Only all others could be the good drivers,I can a good driver.But how does make everyone to be a good driver ?You just can use the product.

    Thumb up 0 Thumb down 2
  7. John A. Bailo says:

    Self drive could also work if we redefine the “car”. For example, many people would ride transit if we could get rid of the last mile problem from the bus or train stop to the cul de sac. For that we could even use slow moving, self driven pedicabs which, even if they went bonkers, would not cause that much harm. And delivery of goods like takeout would be done by a vehicle no more complex than a Segway converted to cargo use.

    Thumb up 0 Thumb down 0
  8. Paul Godsmark says:

    The potential impacts on society could be transformational – on the same scale as the invention of the modern motor car back some 130 years ago, or the invention of the internet. Just think how the car has shaped modern society in North America (and yet is now beginning to choke and suffocate us in some of our cities with pollution and congestion). Fortunately driverless cars could be an important part of the solution to those problems if we plan properly.
    Once these vehicles are certified safe to drive unmanned then we enter a new paradigm which will impact on almost every aspect of our daily lives and will impact most businesses – with some operational and business models being completely disrupted. The scale of the changes are such that we will need clear leadership and policies in place in advance, in order to maximize the benefits and minimize the downsides.
    If you are interested I have a blog dedicated to the impacts of autonomous vehicles where I explain in more detail how things might develop:

    Thumb up 0 Thumb down 0