Former Tesla AI data trainers deliver crucial message on robotaxis
Tesla (TSLA) permabull Cathie Wood is all in on Full Self-Driving (Supervised), calling it the "most impactful AI project out there" while predicting that FSD would drive Tesla to a $2,000-per-share valuation by 2027.
The Ark Invest CEO isn't the only one with high hopes for FSD. Tesla CEO Elon Musk has also touted the tech as the company's most important future-facing project (along with humanoid robots) and Wall Street analysts like Deutsche Bank say investors should focus "more attention... towards the company's robotaxi expansion efforts," and less towards its auto business, which it expects to underperform in 2026.
But according to a new Reuters report, internally, at least some of the people training the AI that powers FSD are skeptical of the tech's viability as the company looks to aggressively expand into new cities in the coming months.
For nearly a decade, Musk has promised that FSD will enable every vehicle the company produces to be fully autonomous. As the class action lawsuits the company is currently fighting show, that vision probably won't be a reality for any Tesla made before 2023.
But even for the next generation of Teslas with all the right hardware, true autonomy may be farther away than many investors think.
Photo by Austin American-Statesman/Hearst Newspapers on Getty Images
Former Tesla AI trainers reveal skepticism about FSD viability
Reuters interviewed nine former Tesla "data labelers" from the company's Utah office, where hundreds of employees scrutinize and label video from the eight external cameras on each Tesla.
The workers train Tesla's AI by labeling traffic incidents as good or bad, identifying items the computer can't recognize, and pinpointing potential trouble spots where the tech needs improvement.
But the problem, according to them, is that FSD is still struggling to execute basic maneuvers, like stopping at a railroad crossing or for school buses, or traveling in the left lane. And despite these consistent issues, Tesla is still publicly touting the tech's prowess while doing big public displays like the Robotaxi launch in Austin last summer.
While FSD and Robotaxi aren't exactly the same technology, according to Tesla, there is apparently a lot of overlap in the work employees are doing to ensure both platforms are safe.
"Inside Tesla, as these events (Robotaxi rollout) approached, staffers worked long hours mapping routes and training the software on specific hazards to make the company's self-driving technology appear more capable than it really is," Reuters reported, citing four former employees.
But the problem is that those fixes the human workers were doing are "impossible to deploy on a broad scale," like the scale Musk has envisioned, where there are millions of autonomous Teslas on the road.
More Tesla FSD news
- Tesla FSD makes dangerous mistake in unfamiliar territory
- Elon Musk's new Tesla update is a joy for customers and investors
- Tesla FSD hits major speedbump with EU regulators
When Waymo goes to a new location, it spends months, or even longer, mapping local roads and hazards to ensure its autonomous vehicles can handle as many real-world situations as possible. And even then, Waymo often gets it wrong in much the same way Tesla FSD/Robotaxi does.
However, Musk has said Tesla is taking an easier route. Instead of doing the labor-intensive work of mapping, Tesla relies solely on cameras and AI to navigate its surroundings.
But according to the Reuters report, humans play a big role in training AI that doesn't know what it is looking at all the time. And even after the AI is trained, it is still prone to making mistakes.
Tesla increases human input as part of FSD AI push
The Utah office employs hundreds of data labelers whom the company needs to train its AI.
In the lead-up to Robotaxi's official public launch last June, despite Musk's insistence on "generalized AI solutions" that didn't require "high-precision maps of a locality," Tesla workers annotated the extensive filming of the limited robotaxi zone to map the area because, apparently, that is the most surefire way to avoid any hiccups in Robotaxi's highly anticipated debut.
In the lead-up to the launch, the staff at the Utah office doubled to about 300 workers, according to Reuters, as they worked primarily on projects "to make the carefully controlled Austin test go smoothly."
But with each update, two employees said that while some driving behaviors improved, others worsened. There was no consistent improvement in the tech, according to one former employee.
Eventually, the human touch powering the AI would be unable, as Tesla rolled out Robotaxi with two sets of safety monitors: one in the vehicle, and one monitoring remotely.
Based on their experiences, four of the former employees said it will take years for Robotaxi to scale up safely. That timeline runs counter to what Musk has promised in the past, like when he predicted that Robotaxi would expand to serve half of the U.S. population by the end of 2025.
Of the 25 vehicles Robotaxi Tracker is currently logging, only three still have human safety monitors in the vehicle, but many have human teleoperators driving them remotely.
Those human teleoperators have come under increased scrutiny, leading Senator Edward Markey (D-MA) to release a report taking AV companies to task for faking their true autonomy.
"Every autonomous-vehicle company refused to disclose how often their AVs require assistance from [remote assistants]-hiding key information from the public about their AV's true level of autonomy," Markey wrote in his report. "This information is critical for lawmakers, regulators, and the public to understand the potential safety risks with AVs."
Related: Autonomous vehicles run into trouble in world's largest AV market
The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.
This story was originally published June 1, 2026 at 8:17 AM.