So now SAE level 3 autonomous cars are "not able to drive in broadest possible situations" like wtf are you even talking about? You are making shit up. I'm sure it's cool. It's just not the only thing out there and its also hard stuck just like every other system even the ones that aren't available which was the point. Its not to detract from tesla its about talking about it plainly from a technology perspective.
And I've given you evidence repeatedly saying they aren't the leader lol. They are a major player and it's cool. Waymo is operating in multiple cities in multiple states on multiple platform. RoboTaxi has like 9 cars in Austin in like 4 zipcodes. They are completely incomparable. I spent an hour in Waymo's yesterday between zilker, north Austin, back to downtown and then east Austin south of the river.
I'm not a Luddite non believer. Lol
Except it's not, the OEM leader. That's the problem. I'm not saying fsd sucks. I'm saying it's not the best or the leader or the most sold or the most used. It's not categorically at the top of anything at the moment, except EV share in the US because of some seriously fucking stupid protrctions against china based EV mfg.
https://www.visualcapitalist.com/the-worlds-top-ev-makers-by-market-share/
This isn't 2022 it's 2026.
Look, Mercedes Benz is the only SAE level 3 car and Tesla isn't. You can keep on with the fanboi shit, but they aren't leading and haven't been for years. Everyone has already at least caught up and many mfg are already surpassing.
You missed my point entirely. I don't think they will and this is like saying LLMs will just get better as time goes on too.
I think things are hitting a hard wall here.
I have a 2018 Mercedes that can do 80% of this. The 2025 Mercedes is on par or better than the Tesla. It's not tesla and I linked you a completely open source thing on par with it as well in any newish whatever car with a sub 1k USD device.
I am convinced that they are at a major major plateau and that its not fixable with more data at this point. They simulate 100x the actual miles driven or something pretty high of a multiple. It's just not improving anymore and I've even experienced some regression in my recent rides.
Its also a wonderful example of how bad of how ineffective it is to dump all this money into AI. Could you imagine how many people you could teach to drive on that many million miles in both a simulator and in real cars with all expenses paid and all they had to do was drive? If would be significantly less than the fleet cost and the training cost. There's a fallacy that once they actually get the solve it'll be worth the money, but I don't see it happening in reality.
Texas isn't that close to end of season Ohio State. Julian saiyan is so much better than game 1. He was fucking in NYC for the Heisman ceremony and it wasn't weird.
Its not close. Georgia, Indiana, Ohio State are on a different level.