Tesla Announces Plan to Achieve Unsupervised Autonomous Driving AI by 2027

On Thursday evening, inside the Warner Bros. Studios lot, Elon Musk rode an autonomous "driverless taxi" on a simulated street, claiming the product would launch "before 2027." The vehicle features neither a steering wheel nor pedals, with Musk seated in the front passenger seat.

Musk refers to this vehicle as the "Cyber Taxi," which aims to achieve a higher level of autonomous driving through artificial intelligence (AI). A video shows occupants in the front seats reading, releasing their hands from the steering wheel, and engaging in various activities—including reading, watching movies, and even sleeping—while the car continues to drive autonomously without their assistance.

To prepare for the market introduction of this driverless taxi, Musk announced plans to "seed" the market next year with an update to Full Self-Driving (FSD), referred to as "Unsupervised FSD"—an upgrade to the current "Supervised FSD."

"Long before that, you will experience driverless taxis," Musk stated during the introduction of the Model 3 and Model Y projects, as well as the Model S and Model X. The Model 3 and Y will achieve unsupervised full autonomy once they receive regulatory approval—in the U.S. first, then internationally—as will the Cybertruck.

Musk mentioned that, pending regulatory approval, Tesla will initially offer unsupervised FSD services in California and Texas.

Although Musk's wording is ambiguous, unsupervised FSD may refer to Level 3 in the National Highway Traffic Safety Administration's (NHTSA) six-tier classification for autonomous driving systems, known as "ADAS."

NHTSA defines Level 3 as "conditional automation," where the vehicle's software handles all aspects of driving, yet a person remains in the driver's seat to take over if necessary. Tesla's current FSD in their vehicles is considered Level 2, meaning humans are responsible for all driving tasks, while ADAS systems assist with specific functions like acceleration.

Some Tesla owners have reported driving several miles without touching the steering wheel. However, these classifications are crucial for regulatory bodies to determine when such features can be rolled out and for Tesla and other manufacturers to claim the achievement of these capabilities.

Wall Street stock analysts monitoring this event have strongly criticized Musk for omitting many details in his driverless FSD commitments.

"Overall, we find Tesla's driverless taxi launch disappointing, with an astonishing lack of details," wrote Tony Sakonaki of Bernstein Research. He pointed out that Musk did not address whether the unsupervised full autonomous driving feature would be backward compatible with Tesla's existing FSD capabilities. The company also did not mention the "path to gaining regulatory approval."

"Tesla has not provided verifiable evidence of progressing toward Level 3 [autonomous driving], nor have they quantified the number of driverless taxis planned," wrote John Coleman, a stock analyst at Jefferies.

Coleman and others also raised a specific technical concern. Musk and his team claim that the driverless taxis will operate solely using onboard cameras, which is simpler than the more common "sensor fusion" approach. In sensor fusion, camera data is combined with lidar and radar systems to create a more comprehensive map of the road.

Compared to using lidar and radar assistance through fusion methods, relying solely on video-based steering may be less reliable for autonomous vehicle navigation.

"Achieving higher levels of autonomous driving solely through visual methods (rather than sensor fusion) is unprecedented," analyst Coleman wrote.

However, Musk emphasized the collective data Tesla vehicles gather and feed into the company's driving simulators. He explained that with one million Tesla cars on the road, it's equivalent to a person experiencing "a million lifetimes," which, he claims, makes autonomous driving "20 to 30 times safer" than human driving.