Advertisement

Tesla keeps touting self-driving but one analyst's near-crash shows it's 'not even close' to autonomy

The interior of a Tesla, displaying Autopilot on a screen.
A Truist Securities analyst took a Tesla Model Y out for a ride to test its Full Self-Driving capabilities.Sjoerd van der Wal/Getty
  • Truist analyst William Stein tested Tesla's Full Self-Driving and said it still lacks full autonomy.

  • Stein said he had to make three interventions, one of which would have led to a "certain accident."

  • The analyst said he is "befuddled" at what Tesla will show at the Robotaxi event.

Following Elon Musk's bold assertions about "solving autonomy," a Truist Securities analyst took a Tesla Model Y out for a ride to test its Full Self-Driving capabilities.

In a note seen by Business Insider, Truist analyst William Stein said the technology was "arguably worse than last time." In its last review, Truist characterized FSD as "stunningly good, but not useful today."

ADVERTISEMENT

"In our opinion the newest version is still stunningly good, but does not "solve" autonomy," the analyst wrote in the note about the latest version. "The shortcomings that we observed make it challenging to imagine what TSLA will reveal in its RoboTaxi event in October."

Stein tested a new Model Y with FSD V12.3.6 with the same start and end points as the last drive in suburban New York. Stein said the weather was clear and dry, which also matched the conditions of the last drive.

Stein flagged Tesla's "demo mode" as one of the surprising features, which is only available during demo drives. Stein said during the first half of the drive that he didn't have to tug on the steering wheel to keep the system engaged. Stein said he even turned his head away from the road to test it further and the system continued for as long as 40 seconds before it issued a warning.

"At 60MPH, this would be a stunning half mile of no-hands and no-eyes autonomy," Stein wrote in the letter, adding that in " this stretch of the highway it seemed safe enough, but in other places, it would have surely caused an accident."

Stein said he also had to make three interventions during the ride. The first one occurred when the Tesla accelerated through an intersection even though the car hadn't completed a right turn yet. Stein said the intervention prevented "an otherwise certain accident."

Stein said the second intervention happened when a police officer used hand motions to ask him to pull to the side of the road to allow other vehicles to pass. The third intervention happened when the vehicle switched lanes despite a solid white line signaling not to change.

Despite these concerning instances, Stein said the experience wasn't all bad. Stein said he was impressed by how well the car adapted to challenging disruptions like lane closures or heavy potholes. He also noted the newer FSD was more active in switching lanes, and the drive "felt more natural overall" than the previous version.

However, he maintained that "solving" may not be the right word for autonomy. Stein added that, "we are still befuddled at what Tesla might show at the Robotaxi event."

Tesla pushed back its historical RoboTaxi event from August 8 to October 10. Musk said in a post on X that he requested "an important design change to the front," and the extra time would help "show off a few other things."

Musk has said Tesla's driverless taxi business will function like "a combination of Airbnb and Uber." He said during the company's second-quarter earnings call that he would be "shocked" if the first Robotaxi ride wasn't possible next year, although he was unable to give an exact timeline. He also admitted he tended to be optimistic when giving estimates.

A spokesperson for Tesla didn't immediately return a request for comment in response to the note.

Read the original article on Business Insider