Elon said:Major improvements are being made to the vision stack every week. Beta button hopefully next month. This is a “march of 9’s” trying to get probability of no injury above 99.999999% of miles for city driving. Production Autopilot is already above that for highway driving.
While perhaps obvious in hindsight, it turns out that the only way to solve self-driving is to solve a big part of real-world AI. That is what Tesla is doing, both in hardware & software. Unless a company is good at AI with tight compute, it’s hopeless. This is insanely hard.
Certainly one of the largest. A company whose name rhymes with Shmoogle is pretty far ahead. But I think we’re the leader in shallow-minded AI haha!
Theory: Main reason for A.I. / autonomy day 2 is a recruitment effort. Tesla will razzle and dazzle about all the hard AI problems they are working on and try to recruit top AI researchers and engineers.
And the SP will go downYes. It will also be a marketing effort like Autonomy Day was. Tesla will razzle and dazzle investors and fans to create a lot of positive buzz. Tesla fans will make videos about how Tesla is the "king of AI" and it is "game over" and Tesla has "won".
Funny you say that - I was just looking at the last 3 quarters - SP went down when the surprise was small negative (Q4), small positive (Q3) and large positive (Q2).Probably. It is weird how the TSLA price often does the opposite of what is expected like going down when Tesla releases a good quarter report.
And yet it still doesn't understand the time-based HOV lanes on I-80, nor the left-side HOV exits.Elon from China A.I. conference July 2020 last year.
Interesting quote: Autopilot works best in California because that is were the engineers are.
From extension to that: autopilot works best in the bay area. I notice that the strongest proponents of FSD on twitter are in the bay area.
Elon said:A major part of real-world AI has to be solved to make unsupervised, generalized full self-driving work, as the entire road system is designed for biological neural nets with optical imagers
Can't even drive down the bay bridge in the eastward direction in the rightmost lane without wobbling all over the place due to uncertainty about the right side of the lane ... just how bad is it outside California lolAnd yet it still doesn't understand the time-based HOV lanes on I-80, nor the left-side HOV exits.
Suspect he is talking about level 4 versus level 5 with the word "generalized". So level 4 maybe, but not level 5 this decade until a major part of AI is solved. Just my 2 cents.Is he saying that generalized FSD is many years away? Doesn't it take a long time to "solve a major part of real-world AI"?
Can't even drive down the bay bridge in the eastward direction in the rightmost lane without wobbling all over the place due to uncertainty about the right side of the lane ... just how bad is it outside California lol