Self-driving without Lidar (what Tesla is doing) seems quite sketchy to me. Tesla seems to have hit some very steep barriers, and most of Musk's timetable predictions have been completely wrong.
Conversation
Replying to
Camera-only self-driving seems like it's just a failed bet at this point. It might not have been quite *wrong* at the time (although I bet against it, and was right), but pretty clearly wrong at this point.
With the benefit of hindsight, it seems kind of amazing that self-driving works even with LIDAR. Apparently, sufficiently high quality telemetry makes it somewhat less AGI-complete?