Tesla’s autonomous driving initiatives, highlighting a sharp divide between corporate goals and operational realities. While Elon Musk continues to promise that unsupervised Full Self-Driving (FSD) will soon achieve superhuman safety, the National Highway Traffic Safety Administration (NHTSA) has escalated a major probe into the system’s failure to handle low-visibility conditions. Regional Robotaxi pilots in Texas are currently facing significant logistical hurdles, including long wait times and a shrinking active fleet despite recent expansions. Furthermore, investigative reports suggest that Tesla’s safety statistics may be methodologically flawed, while European regulators remain skeptical of the technology’s performance on icy roads and at high speeds. Collectively, these documents portray a company aggressively pivoting toward AI and robotics while navigating intense legal, regulatory, and technical challenges.

Why Tesla labelers won't ride robotaxis

Elon Musk Podcast

# Full Transcript

**Podcast:** Elon Musk Podcast

## Participants

- **Stage Zero** (None)

## Transcript

Die Commerzbank steckt voller Geschichten. Und auch wenn wir vielleicht nicht jeden Traum unserer Kunden kennen, wissen wir, wie wir aus ihren Ideen, Plänen und Anliegen gemeinsam Erfolgsgeschichten machen können. Wann sprechen wir über ihre? Gebaut aus 155 Jahren Erfahrung und den Erfolgen unserer Kunden. Commerzbank. Die Bank an ihrer Seite.

7 out of 9 Tesla data labelers, the employees actually hired to review the video footage and train the autonomous driving software, admitted they would not trust the system to drive them. Yeah, and one employee specifically said they wouldn't ride in a company robo-taxi, even if they were paid to do it. Which is just a really stark contrast to the public narrative, you know. The executives insist the system is exponentially safer than a human driver.

Right, they're actively preparing the market for a nationwide robo-taxi rollout. But the people watching the raw feed, the ones going frame by frame categorizing the data, are deeply skeptical of the actual product. So how did the gap between the public safety stats and the physical reality of these cars on the road get so wide? Well, the central safety claim from the executives is that their supervised driving software is up to 10 times safer than a human driver. Which sounds huge on paper.

Right, because internal reports claim there's one major collision for every 5.3 million miles driven on the software. And they compare that to the national average, which is one crash every 660,000 miles. Exactly. But traffic safety researchers looked at that and found a pretty severe methodological mismatch. Because of how they define a crash. Yeah. The company defines a crash as an event severe enough to deploy an airbag or trigger a specific high velocity threshold.

That is an extremely high bar for an accident. It really is. Meanwhile, the national average they're comparing it against includes all police reported incidents. So that covers, you know, minor fender benders that obviously don't deploy airbags. Right. They're basically using two entirely different dictionaries to define the exact same word. Because if you look at the physics of how the company actually measures a crash, they rely on a metric called delta V. Which is what? Exactly. It measures the change in velocity within a 150 millisecond window.

Okay. So the sensors in the bumper and the chassis are looking for a violent, sudden deceleration. Like hitting a brick wall. Exactly. But if you sideswipe a guardrail or, I mean, if you rear end someone at 15 miles an hour and just crush the plastic grill, your forward momentum doesn't stop violently enough in that fraction of a second. So the airbag stays in the steering wheel. Right. And if the physical force doesn't trigger an irreversible restraint system like an airbag, it gets classified differently in their reporting.

But the government baseline...

Why Tesla labelers won't ride robotaxis

Transcript

Built for investment professionals

Every day you wait, your competitors are finding
edge in conversations you're not hearing.

Why Tesla labelers won't ride robotaxis

Transcript

Built for investment professionals

Every day you wait, your competitors are findingedge in conversations you're not hearing.

Every day you wait, your competitors are finding
edge in conversations you're not hearing.