The Misunderstood Self Driving Car

Yesterday I saw a funny cartoon about self driving cars by Scott Adams from the perspective of a robot with an attitude:

When I discuss the impact of automation on the labor market I sometimes run into the argument that self driving cars are still years away and that there are a lot of edge cases to solve before they will replace human drivers.

That argument makes it appear as if there was some distinct moment in the future where we finally say “hurrah, the self driving car is completely finished” and that there will be no impact until then. But that is not how technology progresses.

Instead, technology tends to arrive piecemeal in fits and starts. And it has an impact all along the way already. The same is true for self driving cars. We already have some key components that have entered the market: the most impactful one has been GPS based navigation (which has now come to pretty much every car courtesy of smartphones). While this doesn’t automate the driving per se it does take over a crucial bit which is figuring out how to get from A to B. And that means that cheaper labor can now be substituted for trained labor in driving taxis and limousines!

What will be the next big step? Well one possibility is remote driving. Suppose you have a self driving car but it can’t handle certain edge cases, like getting stumped on a New York city street that’s blocked by an oil delivery truck (we’ve all been there). Instead of having to be able to solve a situation like that completely autonomously one can envision a future of driver pools that stands by whenever a self driving car needs an assist. Here technology provides leverage — a single driver might now be able to “supervise” multiple cars. And of course that driver could be in a low cost part of the country.

This has of course already happened although not yet with cars — the US military has many new drone pilots who have never logged an hour of flight in a real plane. And drones can keep themselves up in the air and takeoff and land by themselves. So even the number of pilot hours relatively to flight hours has already changed. And while I don’t know anything about military pay, in commercial aviation the impact of automation has helped drive down airline pilot pay at the 75% percentile from $150K in 1999 to $100K in 2011.

So let’s please not pretend that we shouldn’t think about labor market consequences of self driving cars because fully autonomous cars en masse are still years away. The impact of what is possible will be felt all along the way.