A PR lie.
Watch a Tesla FSD 12.3.4 video, then one of Mercedes and see for yourself.
Watch a Tesla FSD 12.3.4 video, then one of Mercedes and see for yourself.
Tesla still has this one [1] up where the "not required driver" had to save the vehicle after it ran off the road into fences and etc numerous times [2] [3].
[1]: https://www.tesla.com/autopilot
[2]: https://www.classaction.org/media/matsko-v-tesla-inc-et-al.p...
[3]: https://www.nytimes.com/2021/12/06/technology/tesla-autopilo...
Tesla can not do this no matter what they claim they can do. When their system fails is fails spectacularly and no company in their right mind would want to be liable for that.
It doesn't matter what Tesla can do in 80% of the time when it can't do it 20% of the time.
Consistently running down stationary children in the middle of a empty, well-lit road is not exactly what I am looking for in a autonomous vehicle. In my mind that qualifies as somewhere between criminal negligence and depraved indifference to human life.
It does, however, remain to be seen how the Mercedes system fares in its operating domain, so it is prudent to reserve positive judgement until more evidence emerges.
The continuing problem is that at the end of the day these are just rule-based systems with zero intelligence. They'll work fine 99.9% of the time, and then do something life-threateningly inappropriate in the 0.1% of cases they were not trained to handle.
The only way to make these things as safe as a human driver is to base them on human-level AGI, which is still a long ways off in the future. I don't know if Musk is planning to add his Grok "AI" into the mix, but having a hallucinating chatbot as backup system or driver is not my idea of safety.