Tesla’s Autopilot system doesn’t kill people.* Journalists who write about Tesla’s Autopilot system kill people.
That, at any rate, seemed to be the contention of Tesla CEO Elon Musk in an extraordinary corporate earnings call on Wednesday. The quarterly conference call with investors—a routine affair for most public companies—took a left turn when Musk cut short the Q&A, deeming their questions about Tesla’s finances “boring” and “bonehead” and taking questions instead from a 25-year-old YouTuber who had tweeted at him on Monday. As Tesla’s stock plunged in after-hours trading, Musk detoured into a Trumpian rant—yes, another one—against the dishonest media.
The visionary entrepreneur–turned–press critic blasted journalists for publishing stories about accidents involving Tesla’s autonomous driving software, calling such coverage “outrageous” and “fundamentally misleading.” In fact, Musk suggested, journalists who report on Autopilot crashes are themselves endangering the lives of Tesla drivers by discouraging them from using the software.
“There are 1.2 million automotive deaths per year,” Musk said. “How many do you read about? Basically none of them.” (Note: My own Google News search for “fatal car crash” turned up 2,790,000 results.) “But if it’s an autonomous situation, it’s headline news,” he continued. “And the media fails to mention that actually they shouldn’t really be writing the story. They should be writing about how autonomous cars are safe. But that’s not the story readers want to click on.
“It’s really incredibly irresponsible of any journalists with integrity to write an article that would lead people to believe that autonomy is less safe,” Musk went on. Because when Autopilot users read those stories, “People might actually turn it off and die.”
Lest my journalism cause people to turn off Autopilot and die, I’ve decided I’ll try to follow Musk’s advice in this story. You will not read anything here about a Tesla that may or may not have crashed horrifically while on Autopilot in Mountain View, California, on March 23, hypothetically killing its driver. I also won’t mention that Tesla got kicked off the hypothetical investigation of that noncrash by the National Transportation Safety Board, then claimed that it had withdrawn voluntarily. And I see no need to rehash the fatal nonincident in which a Tesla putatively did not slam broadside into a semitrailer in Florida in May 2016. After all, cars that are not Teslas and are not on Autopilot crash every day.
Just to be safe, I won’t even mention the self-driving Uber that some dishonest media outlets claimed ran over a pedestrian in Arizona in March. I guess it’s possible Musk would be OK with the media reporting on that one, since it didn’t involve his company, but I don’t want to take any chances here. Blood could be on my hands.
Instead I’ll direct you to Musk’s claim that autonomy reduces the probability of a death by 50 percent, even though that claim is impossible to substantiate. I’ll tell you about a 2017 National Highway Traffic Safety Administration report that appeared to say Tesla’s Autosteer feature had made its cars 40 percent safer—a report Tesla has cited ever since as proof that Autopilot saves lives. But I guess I shouldn’t mention that the NHTSA clarified Wednesday that it had never actually assessed Autosteer’s effectiveness. That oft-cited 40 percent figure simply came from a “cursory comparison” of Tesla crash rates before and after Autopilot was implemented, the agency said. Whoops.
I’ll refer you to a story I wrote in 2016 about a man who suffered a pulmonary embolism behind the wheel and drove to the hospital with a big assist from his Tesla. That’s cool, right? As I recall, however, Musk and Tesla’s PR team weren’t happy with even that story, because it went on to discuss in some nuance the broader issues around Autopilot’s safety. Musk and the company were insisting even back then that it would be irresponsible of them not to enable Autopilot in all of their vehicles, given that their internal testing showed it was saving lives.
One problem I probably shouldn’t get into: Tesla has never shared with the public the data that would be needed to evaluate its safety claims. Musk said on Wednesday’s call that the company will start publishing quarterly Autopilot safety statistics in the future, which certainly seems like something it ought to be doing if it wants journalists to be writing all these stories about exactly how safe Autopilot is. So far, Tesla has just been asking us all to take its word for it (and, until Wednesday at least, referring us to that 2017 NHTSA report.)
It’s quite possible Autopilot really is saving lives. But we can’t know that without much more detailed data on how and when it’s used, who’s using it, under what driving conditions, and how its safety stacks up when you control for all those factors. As my colleague Henry Grabar and others have pointed out, comparing crash figures for self-driving cars to those of the general public means comparing the software’s safety to that of 16-year-old boys and drunk drivers, along with the rest. That feels like too low a bar, especially if people are only or mostly using features like Autopilot in fair conditions on well-marked freeways in the daytime while maintaining their lane.
Anyway, the company’s actual earnings report was just fine, at least by Tesla’s recent standards. Tesla is still far behind its own targets for production of the Model 3, the “mass-market” electric sedan that will make or break Tesla’s future. But its sales slightly exceeded expectations, it has enough cash to keep going, and it’s optimistic that dramatic production increases are just around the corner.
Still, Musk quickly soured on questions from investors nervous about the company’s financial situation and the instability of its stock. He seemed to particularly object to questions about why Tesla is losing money on each Model 3 it makes (a “short-term” issue, Musk insisted); whether it’s running out of money and needs to raise more (he says it isn’t and doesn’t); and how long it will be before Tesla reaches the 5,000-per-week production target on the Model 3 that Musk had said last year it would reach by the end of 2017. (Soon! Very soon! Probably as soon as everyone stops asking him about it, OK? Jeez.)
That’s when Musk grew exasperated and turned over the questioning to Gali Russell, an NYU grad student and Tesla retail investor. From the transcript:
Antonio M. Sacconaghi, Sanford C. Bernstein & Co. LLC: And so where specifically will you be in terms of capital requirements?
Elon Reeve Musk, Tesla Inc.: Excuse me. Next. Mourning bonnet questions are not cool. Next?
Operator: Thank you. Our next question comes from Joseph Spak with RBC Capital Markets.
Joseph Spak, RBC Capital Markets LLC: Thank you. The first question is related to the Model 3 reservations, and I was just wondering if you gave us a gauge as maybe some of the impact that the news has had. Like, of the reservations that actually opened and made available to configure, can you let us know, like, what percentage have actually taken the step to configure?
Elon Reeve Musk, Tesla Inc.: We’re going to go to YouTube. Sorry. These questions are so dry. They’re killing me.
Musk proceeded to entertain no fewer than a dozen questions from Russell about a wide range of topics, including the safety of Tesla’s Autopilot system, before giving investors a few last questions and closing the call.
In conclusion, Autopilot is very safe and if you are ever tempted to think otherwise, please go read about some other automobile accidents instead.
There you go, Elon. How’s my integrity now?
*Not technically true.