Skip to Content
Smart cities

Alphabet’s moonshot chief: Regulating driverless cars demands testing the “smarts” of the systems

Astro Teller says Waymo stress-tests its vehicles under “pathological situations,” some involving people in Elmo outfits.
March 27, 2018
Jerey Portje

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.”

 

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

How scientists traced a mysterious covid case back to six toilets

When wastewater surveillance turns into a hunt for a single infected individual, the ethics get tricky.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.