Musk Says Unsupervised Full Self-Driving Is Almost Here, but the Data Says

Tesla’s Full Self-Driving (FSD) has been at the core of its business strategy for years. In fact, the company began selling FSD packages to car buyers as far back as 2019. Buyers who spent thousands of dollars on the feature have been waiting patiently for the company to make good CEO Elon Musk’s lofty promises, but all we’re getting is spin. Musk recently shared a post on X (formerly Twitter) that allegedly shows “exponential improvement,” but it really just shows how far Tesla has to go for full autonomy.

Musk has long promised his electric vehicles are on track to drive themselves, waxing philosophical about cars that can earn money as taxis while the owner is asleep. However, the company recently had to clarify that the FSD it currently offers is not unsupervised—drivers have to be in the car and watching the road, and the system often makes mistakes. Musk has, however, often said unsupervised driving is right around the corner.

On X, Musk reshared a graph showing a substantial improvement in FSD performance. To be clear, this is third-party data crowdsourced from drivers. Tesla has refused to release any of its own data on how well FSD performs. Musk claimed this graph shows exponential improvement, which is not entirely accurate.

The graph (below) does show an improvement between FSD 12.5 and 13.2, with the latter reaching 724 miles between critical disengagements. In this context, a critical disengagement means the FSD system will crash if the user does not intercede. The numbers are certainly getting bigger in FSD 13, but Electrek reports this is thanks to a long overdue update to Tesla’s highway autonomy stack. The company has spent the past two years working intensely on the city driving side of FSD. So, the changes to highway software in FSD 13 look robust because of the update backlog.

But let’s take a step back and look at what these numbers mean. In FSD 13, Tesla vehicles still get themselves into trouble every 724 miles on average. Musk has recently claimed that unsupervised FSD will be available in Q2 2025, but that seems extraordinarily unrealistic. If Tesla could show a pattern of improvements like v12 to v13, perhaps you could make an argument that Tesla is on the right path. However, Tesla doesn’t show any data to anyone—Musk is out there pointing to the crowdsourced data he likes and ignoring everything else.

The fact is, 724 miles between disengagement is nowhere near an unsupervised self-driving system. The NHTSA says to reach that milestone, a car would have to match the average human miles between collisions, which is a whopping 670,000 miles. That hypothetical vehicle would be able to drive itself without getting into multiple accidents during its lifespan, but Tesla is far short of that. It’s a foregone conclusion that Tesla will miss its Q2 2025 timeline for true autonomy, but Musk will probably come up with an excuse and a new timeline, as he is wont to do.

EMEA Tribune is not involved in this news article, it is taken from our partners and or from the News Agencies. Copyright and Credit go to the News Agencies, email news@emeatribune.com Follow our WhatsApp verified Channel