Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

This site may earn commission on affiliate links.
XAI space with Elon happening right now.

Apparently 50 mins ago Elon said they solved FSD, or at least according to the twitterverse. I didn't hear it myself so it's best to go back and listen to the recording.

I've listened to it, and I didn't hear him say that. There was a little mention at about the 44 minute mark that solving FSD is turning out to be easier than they thought it would be, and that they had overcomplicated it, but I don't think he said that it was solved.
 
I've listened to it, and I didn't hear him say that. There was a little mention at about the 44 minute mark that solving FSD is turning out to be easier than they thought it would be, and that they had overcomplicated it, but I don't think he said that it was solved.
Yup, I can't find it either.

However what he said at the 44 min mark is interesting. "I can't speak to what we figured out at Tesla, but in broad terms the answer was way simpler than we thought. We were too dumb to realize how simple the answer was.....so we probably will find out with AGI as well. "

I guess people interpreted this was FSD is solved if Tesla found the answer(in past tense form).
 
Last edited:
About the importance of knowing what’s brewing in a government:

Rivian has paid a lump sum of 85 million euros to VDL for breaking the contract on a possible takeover of VDL Nedcar in The Netherlands.
According to newspaper ‘Eindhovens Dagblad’ (paywalled) this is shown in VDL Groep's annual financial report for 2022.

Excerpt:
- Rivian would take over VDL Nedcar, according to a passage in the financial statements for 2022;
- According to former Nedcar director Paul van Vuuren, there was an agreement between both parties: "There was a signature".
- Rivian broke the contract with VDL after US President Biden came up with a measure to subsidise the purchase of electric cars in the US. Rivian then decided to expand production domestically.
 
Last edited:
To be fair he said "I really consider autonomous driving a solved problem" in June of 2016, yet here we are in July 2023 without it.
Elon has toned it down a lot and has been humbled by Tesla's road to autonomy. He has said many times "turns out to be a lot harder than we expected". This is the first time I hear him say "it's a lot easier than we expected". This is not something anyone would say without the answer...wtf is going on?

Some cocky mathematician can look at a problem and in the middle of solving it can say "I consider this solved"...and then hit a brick wall. Only the guy who has solved it would say "well turns out that was easier than I expected".
 
Reminds me of the artist - musician, painter, author: it matters not - who gets lauded as an "overnight sensation!". They miss that prior decade or two where he labored to get to that overnight.

The image breaks down, however, in that rarely if ever have those artists done anything analogous to someone promising such to thousands who spend upwards of $100,000 for something that has its most important feature "...considered solved".
 
Elon has toned it down a lot and has been humbled by Tesla's road to autonomy. He has said many times "turns out to be a lot harder than we expected". This is the first time I hear him say "it's a lot easier than we expected". This is not something anyone would say without the answer...wtf is going on?

Some cocky mathematician can look at a problem and in the middle of solving it can say "I consider this solved"...and then hit a brick wall. Only the guy who has solved it would say "well turns out that was easier than I expected".
He did like this tweet…so, we will see…. 2 weeks or so?
 

Attachments

  • IMG_6153.jpeg
    IMG_6153.jpeg
    353.6 KB · Views: 136
He did like this tweet…so, we will see…. 2 weeks or so?
If his version of solved FSD is anything like current version of fsd lane changes then fsd is beyond human like. I am still getting lane selection problems but lane changes are god like. The car can squeeze and calculate every chance it gets to over take slower traffic beyond what a normal human can do. Many times I feel FSD has f-ed up and puts me into dicey situations but it manages to recover every freaken time. You guys need to try it to believe it.
 
In the recent spaces for X.ai, Elon mentions that he can't go into specifics, but Tesla has found that the solution to AGI was much simpler than they initially thought. He goes on to stipulate that they'll still need a lot of compute and they're not trying to solve AGI with a laptop, but he reiterates that they've discovered the answer is not as difficult as they thought it would be.

Listening to this I had the feeling they've made some serious advances and with the impending ramp of DOJO, I continue to be excited for Tesla's future.

They have solved FSD, not AGI. And Elon often talks about future things in the present tense once he has convinced himself of a path forward. See every FSD pronouncement of his since 2016.

Elon has a VERY poor record of assessing AI anything. It also seems that Tesla is finally realizing that end to end AI is the way to go for FSD, which is what even George Hotz realized years ago. That it took Elon/Tesla about 7 years to realize this tells you what you need to know about Tesla’s AI acumen.

Look, we all know Tesla is doing well in AI with Dojo, their inference chip, and huge compute cluster, but don’t think they are flawless. Just like with the Model 3 manufacturing ramp, Elon makes mistakes, but then learns and does the right thing in the end.
 
Looks like Tesla is rolling out V4 super chargers with CC contactless payments and touchscreen controls. The touchscreen/payments provider is the Payter Apollo terminal (Payter | Payment terminals | Contactless cashless).

This sounds like a highly premature conclusion to me. All we know is that there are some pix of what might be some relevant hardware. No hint that anything is working now or that anything will be ever.
 
  • Like
Reactions: MikkoJ
I've listened to it, and I didn't hear him say that. There was a little mention at about the 44 minute mark that solving FSD is turning out to be easier than they thought it would be, and that they had overcomplicated it, but I don't think he said that it was solved.
This is so broad that it could just be that “easier” means it’s obviously going to require far less compute than they originally assumed when they designed the FSD computer. Could be that it’s not “solved” but they are finding that the solution is converging on needing significantly less compute than they planned. Less compute = easier.

Could also not be that. But just pointing out it could mean a lot of things and everyone seems to be super excited that it just means robotaxis.
 
Stuff’s going to get really weird. I don’t think traditional investment strategies are going to work well in the near future.

Why?

I recently was directly assigned to be “person who follows AI tools” at my job, paraphrasing. My group is now outcompeting our colleagues/coworkers simply because…we’re working with backhoes while everyone else is still using shovels.

It’s dumber than you can imagine. I show the people using shovels how to use the backhoe and they say “oh. Yeah. Cool. That won’t work for ME though.” And they continue to shovel. I move a mountain in the time it takes them to move an anthill.

Im extremely stupid, though, so maybe I’m wrong about what’s coming, just like I was wrong about growing electrification of the global fleet, China dominance in EVs domestically, and Tesla’s dominance globally. Can’t win em all.