Musk absolutely underestimated the difficulty of full self driving. He's still doing it today. FSD is one of those techs that makes for awesome demos but it has thousands of edge cases that are hard to test for even with millions of vehicle miles recorded in their big AI datacenter.
The tell for me is Hyperloop. That is an optimal environment for FSD, a closed and contained track that has only identical vehicles on it, and yet they still use a human driver.
I only wonder how long it will take before Elon becomes disillusioned, maybe it will never happen. Maybe they'll pull a rabbit out of their hat and actually get it to work. I'm still skeptical.
However I wouldn't d pay attention to the hyperloop thing. It would be trivial to "hardcode" a car to drive at least the tunnel part. It's almost certainly that the FSD team doesn't want to waste time with that.
The tell for me is Hyperloop. That is an optimal environment for FSD, a closed and contained track that has only identical vehicles on it, and yet they still use a human driver.
I only wonder how long it will take before Elon becomes disillusioned, maybe it will never happen. Maybe they'll pull a rabbit out of their hat and actually get it to work. I'm still skeptical.