Alphabet Inc.’s self-driving unit Waymo announced on Monday that it plans to unleash its cars onto the freeway in Phoenix “soon.” The city has long been a test bed for the latest in autonomy, and it’s where Alphabet’s engineers will be the first to send driverless tech hurtling down public roads at high speed without a safety driver.
Good for them. The milestone is a fair reward for Waymo’s more careful approach to autonomous vehicles when compared with its floundering and irresponsible rivals. That Waymo now stands ahead of its competitors should serve as a lesson to companies rushing forward with artificial intelligence: A little caution can pay off.
When Waymo was spun out of Google’s R&D labs in 2016, it was partly in response to itchy senior executives and investors who wondered when the moonshot project would ever start to bring in cash. Google had been a pioneer, picking up self-driving efforts from the US military’s DARPA wing — the department that tries outlandish tech ideas — and injecting it with some Silicon Valley gusto. It found quick and extraordinary success: Jaw-dropping demonstrations showed that the Google car had come extremely far in just a few short years.
I first tried it out in 2015, a throughly boring ride through the leafy residential streets of Mountain View, just around the corner from Google’s headquarters. I say boring because it drove like a grandmother — hesitating at points when most drivers would make a turn or refusing to go around some minor obstacle. Still, the feat at least was tremendously exciting, hinting at a future that was just around the corner. Or so we thought.
The timid driving style of those cars spoke to Waymo’s ethos, a goal that is black and white: Either it would produce a fully self-driving car or it wouldn’t produce anything at all. Any half-measures would put drivers and other road users at risk because its own testing showed frightening evidence that adding piecemeal autonomy meant drivers have a tendency to become complacent and tune out.
Elon Musk, naturally, held a different view. In 2014, Tesla Inc. began rolling out the irresponsibly named Autopilot — which was neither auto nor a pilot. Instead, it is a limited function that can only truly be relied upon on freeways and never without human drivers who have their hands on the wheel and are paying attention. “Systems requiring human driver oversight are not self-driving and should not be called self-driving,” a coalition of car companies, including Waymo, warned at the time. A Washington Post investigation in June 2023 logged 17 deaths and 736 crashes in the US in which Autopilot had been a factor.
Enjoying the stock bounce and publicity that came with it, Musk was able to claim he was at the forefront of technological innovation, saying the hardware in existing Tesla cars would be able to achieve the kind of full self-driving that Waymo was striving for. Only recently have some of these mischaracterizations started to come home to roost: In December, a recall of more than 2 million Teslas demanded a more robust safety system to make sure drivers weren’t using Autopilot incorrectly.
Elsewhere, it was also full steam ahead at Cruise, the autonomous-driving company majority owned by General Motors Co. In something of a coup, in December 2020 the company beat Waymo to being the first to roll out self-driving cars without a safety driver, in San Francisco. (In an area that included my old street, as it happened. Nightly, aggravated humans would swear and honk at the little cars. When not on the road, the cars gathered in the parking lot of a bank, shuffling about like Daleks.)
“We’re on a trajectory that most businesses dream of, which is exponential growth,” Kyle Vogt, Cruise’s chief executive officer at the time, told investors. By the end of 2022, the company was testing in three US cities. “Imagine where we’ll be in a year!” Vogt wrote on Twitter. Out of a job, as it turned out: Vogt departed after a Cruise car was involved in an accident that left a woman critically injured. Actually, it would be more accurate to say Vogt left after authorities revoked the company’s testing permit after they determined it had withheld some footage of the accident. The newly uncovered video showed the Cruise car dragged the woman 20 feet at 7 miles an hour before coming to a stop — on top of her. It’s not the only significant accident involving a self-driving car. In 2018, an Uber test vehicle, with a distracted human driver, struck and killed a woman in Tempe, Arizona.
Such examples speak to the frenetic rush to get technology out on the roads and into company earnings calls. For sure, Waymo’s own testing in San Francisco has not been without its frustrations. But, as one academic described it to NPR, Cruise had been the “bull in a china shop” when it came to rolling out the technology as quickly as possible, while Uber’s self-driving program, whistleblowers said, had rushed to address what former CEO Travis Kalanick considered an existential threat to his business. (Uber is now a partner with Waymo, though not until after settling a court case in which it was accused of stealing Google’s technology in its efforts to catch up.)
According to the most recently available data, Waymo’s cars have traveled 7.1 million miles and have caused just three minor injuries. Waymo has consistently acted more slowly despite considerable pressure from outside investors wondering why, with all of its financial might, it wasn’t running rings around those competitors. Well, today it is — a factor worth considering in Google’s AI battle against Microsoft and OpenAI, in which the company is again facing criticism for being too slow and cautious. Waymo is surely a reminder that, when dealing with extremely powerful technology, patience is a virtue — not just for the people who come into contact with it, but for the company’s long-term success.
A message from Advisor Perspectives and VettaFi: To learn more about this and other topics, check out our podcasts.
Bloomberg News provided this article. For more articles like this please visit
bloomberg.com.
Read more articles by Dave Lee
