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

[SOULPEDL]

Im driving around in Costa Rica for a few weeks... pos ice rental! FSD here would need HW4 IMO. Prioritize pot-hole detection.

 

[lklundin]

Maybe, but I wouldn't have expect such from Fridman. I gather he's more of human-computer AI researcher. From MIT page: “Human-Centered Artificial Intelligence,” which is defined by two goals: (1) the AI system must continually improve by learning from humans while (2) creating an effective and fulfilling human-robot interaction experience. "

The nuts and bolts of autonomous driving equations may be outside of his wheelhouse.

I think that's a helpful observation.

I also think both Tesla and their customers need to have a good understanding of how humans interact with Tesla's FSD system - until the day when the driver no longer needs to be in the loop.

 

[sundaymorning]

'teslas are appreciating assets'! EM in the podcast.
If they took that seriously leasing on their terms is very profitable. They could loan them out for free and take them back years later at a profit.

He’s not talking monetary, he’s talking technology.

 

[kbM3]

Did it occur to anyone else how extraordinary it is that Elon can handle himself so well in an interview with an extremely knowledgeable professor in AI from MIT?

Can you imagine any other automotive CEO performing like this?

 

[Cobbler]

Did it occur to anyone else how extraordinary it is that Elon can handle himself so well in an interview with an extremely knowledgeable professor in AI from MIT?

Can you imagine any other automotive CEO performing like this?

Can you imagine Bob Lutz sitting in that chair?

 

[electrichapp]

What website did you get this from? I'd like to bookmark this

Opricot Open Interest|Volume|Max Pain

 

[M3Rider]

To get to 99%, it may be enough to identify pedestrians, wheels, plastic bags, roadkill and potholes.
To get to 99.99%, there may be hundreds of different road hazards, going up to 100’s of thousands as you march to 99.999%.

Am I missing something. Does someone know how Tesla can handle edge case explosion?

They have to be only twice as good as a human, not 1000 times better than a human. Accidents rates will attest to that. While they will be improving going forward, they don't need all the 9s in the world to make it official.

My opinion is now it might be the last chance for anyone who understands the implications of this to front run wall street bean counters.
Soon people would wake up and realized they need to apply Waymo valuation to Tesla, and probably multiple times over and over.

Yeah, the risk is by how much(how many years) they miss the goal. What we see is aspirations. Market will want a proof in the form of regulator's approval and nothing extra bad happening afterwards.

Taking those video frames and turning them into consistently safe and correct car control actions, on the other hand, will require quite a bit of compute power(nobody really knows for sure how much, yet). Running networks capable of that on low resolution or low frame rate may be far more compute hungry than running their current networks on full res/full frame rate.

They seem to have high confidence in HW3 being able to handle it, and Karpathy, in particular, is someone I'd trust to best make that judgement, but just taking their current performance at face value doesn't necessarily guarantee that.

Elon stated he currently has the FSD that stops for red lights. I would assume the 10% usage is given for the real Beta FSD, not for AP.

 

[Sean Wagner]

Did it occur to anyone else how extraordinary it is that Elon can handle himself so well in an interview with an extremely knowledgeable professor in AI from MIT?

Can you imagine any other automotive CEO performing like this?

Indeed. This is one of the chief reasons Tesla can take decisions on self-driving hardware for hundreds of thousands of cars way in advance. The downside was a bunch of derelict Kuka robots and many overworked programmers flown in from all over the world improvising the solutions that ultimately saved Tesla. Let's not forget all these people's contributions [cum very nice remuneration, I hope]. Um. Might the SEC be a joke by entities outside the simulation?

Edit: that was very much worth watching despite the references to a soon bygone era of human-machine supervision. Next comes...

 

[generalenthu]

I am still bummed about the magnitude of EU incentives available in 2020 and beyond assuming there are no last minute change to rules. Here is the quick table I was able to pull together. I keep telling myself, when something is too good to be true it normally is. I see @Prunesquallor has a lower estimate. Not sure why. I think its reasonable to assume that tesla will get at least 2/3 of the benefit.

 

[HG Wells]

I am still bummed about the magnitude of EU incentives available in 2020 and beyond assuming there are no last minute change to rules. Here is the quick table I was able to pull together. I keep telling myself, when something is too good to be true it normally is. I see @Prunesquallor has a lower estimate. Not sure why. I think its reasonable to assume that tesla will get at least 2/3 of the benefit.

View attachment 396221

Starting January 1 2020 Tesla should ship all and I mean ALL cars to Europe. Not one should go anywhere else.
This is too good to pass up.

 

[Doggydogworld]

They have to be only twice as good as a human, not 1000 times better than a human.

I think that's delusional. Here in Texas lawyers run "Hit By a Company Vehicle?" ads because awards can be 10-100x higher for the same injury. No company can afford to operate a robotaxi service without a liability shield. And lawmakers won't create liability shields for a 2x improvement. Especially not a 2x Muskstat. You'll need something like 10x verified by independent 3rd party, at a minimum.

 

[Nocturnal]

Has Tesla talked at all about FSD cars being able to talk to each other? Perhaps that is next level but FSD cars cooperating with each other can be dramatic in how our transportation works. I don't recall the numbers, but I watched an excellent video on the subect and a surprisingly low % of FSD cars on our highways can make dramatic improvements in throughput by enforcing rational acceleration, braking, lane changes...

I think that's delusional. Here in Texas lawyers run "Hit By a Company Vehicle?" ads because awards can be 10-100x higher for the same injury. No company can afford to operate a robotaxi service without a liability shield. And lawmakers won't create liability shields for a 2x improvement. Especially not a 2x Muskstat. You'll need something like 10x verified by independent 3rd party, at a minimum.

Humans drive like garbage so 10x shouldn't be that much harder. Even if you just eliminated ALL impaired driving that would get you part of the way there.

44% of all traffic deaths are caused by drugs or alcohol. Eliminating that risk is a 2x improvement right there.
Impaired Driving: Get the Facts | Motor Vehicle Safety | CDC Injury Center

Driving while being overly tired definitely get's rid of another sizable chunk.
Dangers of Drowsy Driving

 

[MarcusMaximus]

Elon stated he currently has the FSD that stops for red lights. I would assume the 10% usage is given for the real Beta FSD, not for AP.

Right, with unknown accuracy, and no word on correctly making turns at intersections, being able to find parking spots, etc. Improvements, even just to accuracy, may/likely will require architectural changes to increase network capacity. That’ll take more compute power.

 

[Sean Wagner]

So that’s a no? In which case Tesla needs an alternative location. If a European country won’t move fast enough for Tesla, Elon will do something different. What are the most logical possibilities?

I think @Electric Dream was speaking for the UK. Here's a counterintuitive idea - while Czechia and Slovakia basically have full employment thanks to their excellent, somewhat cheaper workforce and the German automotive industries' many factories [also some development centers] including the very large supplier business, Tesla could be an attractive place to work. Especially when the EV disruption begins to be felt there too. IT talent is also available.

 

[printf42]

OT,
It’s weekend again right? Seems Market Maker succeed at pining the SP to Max Pain once more.

Another thought from the Lex interview.
Seems Elon’s worries on AGI has another layer.
That we are in a simulation, and AGI would be able to find the proof of that. Then it would trigger the simulators to stop/restart our instance. Another scenario how AGI would end the world as we know it.

 

[Nocturnal]

OT,
It’s weekend again right? Seems Market Maker succeed at pining the SP to Max Pain once more.

Another thought from the Lex interview.
Seems Elon’s worries on AGI has another layer.
That we are in a simulation, and AGI would be able to find the proof of that. Then it would trigger the simulators to stop/restart our instance. Another scenario how AGI would end the world as we know it.

BRB, I need to go buy some cool shades and leather trench coats.

 

[Sean Wagner]

Humans drive like garbage so 10x shouldn't be that much harder.

The problem is that in our self-assessment, we diverge massively from the norm and are drivers above reproof.

 

[Nocturnal]

The problem is that in our self-assessment, we diverge massively from the norm and are drivers above reproof.

At some point driving your own car will be crazy expensive. If we live in a world where self driving is pervasive and extremely safe, if you get into an accident and kill somebody the lawsuit will be crazy. "You could have avoided this if you had used safe self driving". Insurance rates will skyrocket for self drivers.

 

[printf42]

OT,
It’s weekend again right? Seems Market Maker succeed at pining the SP to Max Pain once more.

Another thought from the Lex interview.
Seems Elon’s worries on AGI has another layer.
That we are in a simulation, and AGI would be able to find the proof of that. Then it would trigger the simulators to stop/restart our instance. Another scenario how AGI would end the world as we know it.

And, to anyone who doesn’t believe we are in a simulation, it would appear like AGI had a breakthrough on fundamental physics, then an accident in LHC while experimenting that theory wiped out our universe in a blink.

 

[kbM3]

I am still bummed about the magnitude of EU incentives available in 2020 and beyond assuming there are no last minute change to rules. Here is the quick table I was able to pull together. I keep telling myself, when something is too good to be true it normally is. I see @Prunesquallor has a lower estimate. Not sure why. I think its reasonable to assume that tesla will get at least 2/3 of the benefit.

View attachment 396221

Congrats on being quoted so much in Rob Maurer’s TechCastDaily podcast on this subject

Right, with unknown accuracy, and no word on correctly making turns at intersections, being able to find parking spots, etc. Improvements, even just to accuracy, may/likely will require architectural changes to increase network capacity. That’ll take more compute power.

I believe he quoted 98% red light detection in the q4 cc where they need to get to 99.999%. Which understates the magnitude of improving the error rate of 1 in 50 by a factor of 2,000.