Yeah, that's basically the opposite of what I'm saying. The problem is enormously difficult, but it won't be solved as a problem in itself, it will suddenly become solved as a side effect of better AI, exactly like numerous cognition-related problems got suddenly solved with the advent of LLMs. And we've seen that there are many players working on better AI, and the main ones all maybe one/ two years away from each other. Self driving will become commoditized much before any of the early players can make substantial amounts of money out of it. Compared to LLMs, it's even worse, as self-driving can't really get superhuman and people will happily settle for whatever works acceptably well (where acceptably means at the same level as a good human driver).