In January’s Tesla earnings call, an analyst asked Elon Musk about his belief that Tesla would achieve Level 5 autonomy—jargon for a car that can drive itself in all situations—by the end of the year.
“I’m confident based on my understanding of the technical roadmap and the progress that we’re making between each beta iteration,” Musk said.
But six weeks later, Tesla’s director of Autopilot software, CJ Moore, contradicted Musk in a March meeting with California regulators. That’s according to a memo obtained by transparency site Plainsite via a freedom of information request.
According to LinkedIn, Moore has been at Tesla for almost seven years and became director of Autopilot software in 2019. He was one of four Tesla representatives who participated in a conference call with officials from California’s Department of Motor Vehicles on March 9.
“DMV asked CJ to address, from an engineering perspective, Elon’s messaging about L5 capability by the end of the year,” reads a memo by a DMV official summarizing the meeting. “Elon’s tweet does not match engineering reality per CJ.“
Someone—presumably the California DMV—appears to have tried and failed to redact that last sentence. That part of the letter appears as white space in the PDF, but it can be copied and pasted into another document.
It’s not clear what tweet the memo is referring to. I looked at Musk tweets in the weeks before March 9 and didn’t find any about the timeline for achieving level 5 autonomy.
But January’s earnings call wasn’t the only time Musk has predicted that Tesla would achieve level 5 autonomy by the end of 2021. In a December 2020 interview, Musk said he was “extremely confident” that Tesla vehicles would reach level 5 within a year.
The DMV memo then continues to summarize the Tesla representatives’ comments:
Tesla is at Level 2 currently. The ratio of driver interaction would need to be in the magnitude of 1 or 2 million miles per driver interaction to move into higher levels of automation. Tesla indicated that Elon is extrapolating on the rates of improvement when speaking about L5 capabilities. Tesla couldn’t say if the rate of improvement would make it to L5 by end of calendar year.
It’s not clear how Tesla defines a “driver interaction,” but this appears to be an extremely high bar. When I last took a close look at videos of the FSD beta in real-world situations, I saw the software make a number of errors in just a few hours of video. Some recent videos have continued to show frequent mistakes, so it seems unlikely that Tesla is going to drive down the rate of mistakes to once every million miles over the next eight months.
This isn’t the first time Tesla has appeared to tell California regulators one thing while Musk told the public another. Letters to the California DMV in November and December argued that the FSD beta “will continue to be an SAE Level 2, advanced driver-assistance feature,” not a fully autonomous system. Tesla indicated that work on “true autonomous features” would start at a later date.
FSD has been one to two years away for five years
All of this should feel familiar to longtime Tesla watchers. Two years ago, Elon Musk confidently predicted that by the end of 2020, Tesla’s software would be sophisticated enough to allow Tesla vehicles to operate as fully autonomous robotaxis. Needless to say, nothing like that happened.
Three years before that, in June 2016, Elon Musk was asked how long it would take to achieve fully autonomous vehicles.
“I really would consider autonomous driving to be basically a solved problem,” Musk said. “I think we’re basically less than two years away from complete autonomy.”
Four months later, Tesla unveiled a new version of its Autopilot hardware. The company billed the new vehicles as capable of full self-driving and began taking payments for Tesla’s future full self-driving software. Musk predicted that a Tesla would be able to drive from coast to coast “without the need for a single touch” by the end of 2017.
CJ Moore wasn’t the first Autopilot leader who was uncomfortable with his boss’s grandiose promises about full self-driving technology, either. Sterling Anderson was the head of Tesla’s Autopilot program at the time Tesla began selling the full self-driving package in 2016.
“The announcement shook up some engineers, because they believed that the product that was released wasn’t designed to be self-driving,” according to a 2017 story in the The Wall Street Journal. When an employee asked Anderson about the move, he reportedly responded that “this was Elon’s decision.”
Anderson quit soon afterwards and wound up co-founding the self-driving startup Aurora. Musk’s decision to market Autopilot as a “full self-driving” system “was a factor in the decision by Mr. Anderson and at least two other engineers to leave the company,” according to the Journal.