Try to imagine a future with self-driving cars. What do you picture? Are you sliding into your own Tesla Model S, or are you calling up Driverless Cars Company X for a ride? Do the cars circle campuses and downtown streets until summoned? Or do they quietly return to driveways and parking lots, ready to be woken up when needed? For all of Elon Musk’s confidence, it is still unclear how self-driving cars will fit into our society.
“It’s a bit of a morbid interest,” laughed Stilgoe, “But people like me are extremely interested in accidents because they show the reality of technology, not just the shiny public image.”Jack Stilgoe, visiting professor from the University College of London, became increasingly interested in self-driving cars after one crashed in 2016, resulting in the driver’s death and reawakening some doubts about the technology.
Stilgoe is visiting the CIRES Center for Science and Technology Policy Research (CSTPR) for a year to research how driverless cars are developed, governed, and perceived by the public. He’s especially interested in machine learning, a type of artificial intelligence that provides computers with the ability to learn without being explicitly programmed.
“I’m interested in the novel aspects of the science of self-driving cars, and how they relate to machine learning and artificial intelligence,” said Stilgoe. “This is the particular thing that has enabled self-driving cars to suddenly go from seeming completely impossible, about 10 years ago, to now seeming sort of inevitable.”
But, Stilgoe said, as with all exciting emerging technologies, there are questions we should all ask about how self-driving cars are emerging and whose interests they serve. For example, what are issues that should be talked about in public debates? And who should we, the public, trust to tell us the truth?
Stilgoe pointed to some past examples of exciting technological advancements we can draw lessons from. The emergence of cars—normal, driver-required cars—is a good analogy to the impacts that self-driving cars might have.
“When cars emerged at the start of the twentieth century, they radically reshaped social norms and the structure and fabric of our cities, in ways that people didn’t anticipate at the time,” said Stilgoe. “I think we need to do better at anticipating the impact of self-driving cars, because the promises are just as big as they were for regular cars back in the 1900s.”
For example, cars drastically changed how we build public areas. Think of Boston and San Francisco—cities that were built before cars were as commonplace as they are today—compared to cities like Los Angeles or Houston, which were built with cars in mind and have the parking lots to prove it. Self-driving cars could have an impact that is equally dramatic on our cityscapes.
Stilgoe also referred to agriculture biotechnology, which many expected would revolutionize the food system. In various ways it did, but not all of the claimed benefits came to fruition, and many people were skeptical of the benefits that were touted by agriculture companies. Stilgoe makes the point that not all of the claims of people and companies touting self-driving cars should be taken at face value.
Stilgoe is particularly interested in some of the different directions that widespread adoption of driverless cars could take in the future. He believes that the philosophy and design of machine learning algorithms will shape the future one way or another.
“Self-driving cars are seen by some engineers as just like a game of chess, with a machine learning to do it as well as or even better than humans,” explained Stilgoe. “That leads you to a hubristic model, where you say that anything that the world can throw at me, I can navigate as a self-driving car.”
He juxtaposes this with a model that assumes the self-driving cars are not good at reacting to unexpected events, leading to a future that has separate routes for self-driving cars, or a future that requires “smart roads.”
Both of these models of the future raise philosophical and political questions, many of which are being fleshed out in government and (more loudly) in the media. For example, should we allow self-driving cars on the road with normal cars, or will that cause too many accidents? For that matter, how many accidents involving a self-driving car is too many? You can probably imagine many more questions, and it’s clear that there’s a lot about our future with driverless cars that we can’t predict quite yet. But if there’s one thing we do know about self-driving cars, it’s that they will be in our future one way or another.
By Alison Gilchrist