The head of Alphabet’s X division says that society needs to develop new ways to regulate autonomous vehicles, ideally by testing the “smartness” of specific systems.
Astro Teller, X’s chief of moonshots, made the point during an onstage interview at MIT Technology Review’s EmTech Digital conference in San Francisco. He was responding to a question from senior editor Martin Giles, who asked whether companies should halt testing of driverless cars on public roads following the fatal accident involving an Uber autonomous car earlier this month (see “What Uber’s fatal accident could mean for the autonomous-car industry”).
Teller declined to discuss Uber specifically, but he said he’s intentionally stepped in front of moving Waymo cars “many times” himself, in a clear suggestion that he believes the vehicles first developed by his own X division are safer.
“I feel incredibly comfortable doing that, because I know what testing we’ve done,” he said.
Indeed, in response to an earlier audience question, Teller stressed that developing and evaluating safe and predictable AI systems requires incredible amounts of real-world and virtual testing. He said that Waymo has driven some 2.5 billion miles in simulations that closely match real-world conditions in the last 12 months, while its cars have driven more than five million miles on physical roads with “hyper-well-trained drivers behind the wheel.”
Moreover, he added that for many thousands of those real-world miles, they deliberately subjected the system to “pathological situations," testing the sensors and software in unpredictable scenarios. They've had people dressed as Elmo, hiding in bags, and lying down on skateboards all pop out in front of the vehicles, Teller said.
“You have to test in the physical world, you have to do it incredibly carefully, and you have to create the worst situations you can while you’re doing it in very controlled circumstances,” he said.
Waymo’s chief executive, John Krafcik, was more explicit about the relative safety of its system versus Uber’s in recent days. During a speech at the National Automobile Dealers Association conference in Las Vegas, he said of the accident, according to Bloomberg: “We have a lot of confidence that our technology would be robust and would be able to handle situations like that one.”
The hype around DeepMind’s new AI model misses what’s actually cool about it
Some worry that the chatter about these tools is doing the whole field a disservice.
These materials were meant to revolutionize the solar industry. Why hasn’t it happened?
Perovskites are promising, but real-world conditions have held them back.
Why China is still obsessed with disinfecting everything
Most public health bodies dealing with covid have long since moved on from the idea of surface transmission. China’s didn’t—and that helps it control the narrative about the disease’s origins and danger.
A quick guide to the most important AI law you’ve never heard of
The European Union is planning new legislation aimed at curbing the worst harms associated with artificial intelligence.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.