Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
This site may earn commission on affiliate links.
With an ensemble of neural networks cross checking each other, and the power of large language models doing some reasoning I think we can get there, although current hardware is insufficient, for what I have in mind.
I agree, and that's the real type of engineering that Waymo spent the last ten years doing (apart from developing their own in-house sensors and compute). I seriously doubt Tesla will get to the equivalent ODD as Waymo (ie downtown to highway) this decade even if they start the pivot from driver assisted to autonomous today. But let's hope they do!
 
Last edited:
I think some people have set the bar too high.....it’s either Data from Star Trek or nothing....my bar is, can the car drive better in most circumstances than most drivers....we are still not there...but the writing is on the wall
Worth noting that in some ways, we have already surpassed Data. Recall, in early seasons Data didn't understand humor and didn't have emotions like empathy.
 
Somehow this very important FSD Beta Tweet didn't make it into this thread. Maybe I missed it? (I looked!) People really falling down on the job.

Elon Musk, March 29th, 2024:
"Most people still have no idea how crushingly good Tesla FSD will get. It will be superhuman to such a degree that it will seem strange in the future that humans drove cars, even while exhausted and drunk! Cars will take you where you want automatically, just like getting in an elevator and pressing a button, something that also used to be manual."


I guess in retrospect we know now that he's talking about FSD, not FSD Supervised. So sky's the limit for FSD! I think this is consistent with @Daniel in SD's claim that FSD is L5, though I really can't keep track.

And I guess there can't be any more FSD Beta Tweets, since FSD v12 is not Beta. Fortunately Twitter still exists, so there can still be Tweets. But not sure about this thread. Maybe it needs a new title.
 
Somehow this very important FSD Beta Tweet didn't make it into this thread. Maybe I missed it? (I looked!) People really falling down on the job.

Elon Musk, March 29th, 2024:
"Most people still have no idea how crushingly good Tesla FSD will get. It will be superhuman to such a degree that it will seem strange in the future that humans drove cars, even while exhausted and drunk! Cars will take you where you want automatically, just like getting in an elevator and pressing a button, something that also used to be manual."


I guess in retrospect we know now that he's talking about FSD, not FSD Supervised. So sky's the limit for FSD! I think this is consistent with @Daniel in SD's claim that FSD is L5, though I really can't keep track.

And I guess there can't be any more FSD Beta Tweets, since FSD v12 is not Beta. Fortunately Twitter still exists, so there can still be Tweets. But not sure about this thread. Maybe it needs a new title.
This is a typical Elon tweet.

He could as well has said that ”people will have no idea how crushingly good VTOL/time travel/cold fusion/robots will get”.

Elon takes a thing that people believe will exist in 5-10 years and inserts himself and Tesla into that narrative.

As for FSD, we’re still as far from Tesla robotaxi as we were in 2019 tbh. Still science-fiction. FSD (beta, err supervised) is impressive is some ways. But also mostly a tech demo that lacks a real value proposition until they remove the driver. 90% of the work is left for that. The so called march of nines.
 
Last edited:
This is a typical Elon tweet.

He could as well has said that ”people will have no idea how crushingly good VTOL/time travel/cold fusion/robots will get”.

Elon takes a thing that people believe will exist in 5-10 years and inserts himself and Tesla into that narrative.
He’s selling the future....and he has an impressive track record
 
He’s selling the future....and he has an impressive track record
Impressive how? At selling, yes. In practice FSD next year for ten years, and now he’s peddling robots that won’t be a thing for 15 years.

Making electric cars is cool, but that didn’t really require any major invention. Just massive risk taking and a lot of guts. SpaceX is impressive, but I’m taking about Tesla now.
 
Impressive how? At selling, yes. In practice FSD next year for ten years, and now he’s peddling robots that won’t be a thing for 15 years.

Making electric cars is cool, but that didn’t really require any major invention. Just massive risk taking and a lot of guts. SpaceX is impressive, but I’m taking about Tesla now.
Look, if you want to know what science will look like in the future...you get a quote from someone who was in the Big Bang Theory
 
I'm guessing he will be about 20 years off on average for guessing computer vision only autonomy next year every year since then when it arrives for city+highway around 2038-2039 if ever.
He has a useable product....which you can use while it improves....and as a project it may never finish....very little technology stops because it’s reached it’s pinnacle....even sailing boats are constantly improved
 
I guess in retrospect we know now that he's talking about FSD, not FSD Supervised. So sky's the limit for FSD! I think this is consistent with @Daniel in SD's claim that FSD is L5, though I really can't keep track.

And I guess there can't be any more FSD Beta Tweets, since FSD v12 is not Beta. Fortunately Twitter still exists, so there can still be Tweets. But not sure about this thread. Maybe it needs a new title.
Clearly he's talking about a new FSD beta which is definitely not FSD (supervised) so it's still an FSD beta tweet.
 
Honestly, I think V12 with ASS/Banish, a few more rounds of input data, and a few tweaks (e.g. don't take off like a rocket from every stop sign, etc) is already pretty close to an L3/4 candidate if they want to go there. Maybe not L5 yet. It doesn't always do things the optimal way, or the way I'd like it to do them, but I bet it can safely get me somewhere while I read a book, with some basic fallback with a few seconds notice to take over if it gets truly confused/stuck.
 
Honestly, I think V12 with ASS/Banish, a few more rounds of input data, and a few tweaks (e.g. don't take off like a rocket from every stop sign, etc) is already pretty close to an L3/4 candidate if they want to go there. Maybe not L5 yet. It doesn't always do things the optimal way, or the way I'd like it to do them, but I bet it can safely get me somewhere while I read a book, with some basic fallback with a few seconds notice to take over if it gets truly confused/stuck.
Are you aware that you need at a minimum 10000 miles per disengagement to leave L2? Probably a lot more.
 
  • Like
Reactions: Ben W
Meanwhile:
"Last month, Tesla machine-learning scientist Ethan Knight became the fourth Tesla engineer to leave the car company for xAI, the startup’s website shows. At Tesla, Knight had overseen the team working on computer vision for Tesla’s self-driving technology, according to a former employee. His departure follows those of engineers who worked at Tesla on supercomputing, Autopilot and artificial intelligence infrastructure and who have joined xAI in the past 12 months."

https://www.theinformation.com/articles/musks-xai-is-poaching-engineers-from-tesla (paywalled)
 
  • Funny
Reactions: Daniel in SD
Meanwhile:
"Last month, Tesla machine-learning scientist Ethan Knight became the fourth Tesla engineer to leave the car company for xAI, the startup’s website shows. At Tesla, Knight had overseen the team working on computer vision for Tesla’s self-driving technology, according to a former employee. His departure follows those of engineers who worked at Tesla on supercomputing, Autopilot and artificial intelligence infrastructure and who have joined xAI in the past 12 months."

https://www.theinformation.com/articles/musks-xai-is-poaching-engineers-from-tesla (paywalled)
Can’t say we weren’t warned. 🤷‍♂️
 
  • Like
Reactions: spacecoin
Are you aware that you need at a minimum 10000 miles per disengagement to leave L2? Probably a lot more.
Where does that figure come from? Most disengagements aren't safety-critical anyways, in my experience. Mostly comfort issues, not getting honked at, or missing a turn which would then add travel time on re-route. None of those things really have to be perfect to get the job done in a basic sense.