3 comments

[ 5.0 ms ] story [ 18.3 ms ] thread
Tesla's FSD is fully dependent on visual camera feedback.

When (not if) visuals are obscured, there is no backup. The vehicle is driving blind.

This is a fundamental engineering failure --- a lack of redundancy in a critical safety system. It can't/won't be fully overcome without additional feedback in my opinion.

I've read that going without sleep for just 24 hours is worse for your reaction times than having a blood alcohol content of 0.08, exceeding the legal limit.

Obviously the most responsible choice is to not drive while sleep deprived, or to pull over to a safe location and rest if one is too drowsy to operate a vehicle.

That said, if one is refusing to do that, for whatever reason, I have a hard time believing the upcoming FSD 14 stack is going to be more dangerous than driving drunk, or than driving while so drowsy as to be equally unsafe to driving drunk.

The older pre-NN stacks definitely weren't "there", but it requires either a meaningful ignorance or motivated partisanship to argue that FSD 13 isn't at least approaching the threshold of being a better driver than the worst drivers on the road, if not definitively better.