Poll: When will FSD V12 be in wide release?

[dbldwn02]

Tesla FSD Version 12 — Not As Good As Hoped & Planned? - CleanTechnica

Sign up for daily news updates from CleanTechnica on email. Or follow us on Google News! Tesla Full Self Driving (FSD) Version 12 has started rolling out to customers. There are, naturally, some owners raving about its improvements on social media. The update uses an end-to-end neural network to...

cleantechnica.com

Instead of paying $12K and 'hoping', I think the smart money would be to wait for FSD to actually work THEN buy a new Tesla with it.

I agree. Keep in mind, If it ever goes level 4, it's going to be way more than $12k.

Better to maybe buy the dip right now and if it goes level 4, stock sky rockets and then you can afford the $15k+ FSD.

 

[uscbucsfan]

I agree. Keep in mind, If it ever goes level 4, it's going to be way more than $12k.

Better to maybe buy the dip right now and if it goes level 4, stock sky rockets and then you can afford the $15k+ FSD.

Maybe, maybe not. I don't know that it ever will be level 4, but if it is, what's the demand threshold? $15k may be too high. The high sales price is, IMO, to push more to a subscription as we all know the take rate on FSD is low.

 

[E90alex]

I agree. Keep in mind, If it ever goes level 4, it's going to be way more than $12k.

Better to maybe buy the dip right now and if it goes level 4, stock sky rockets and then you can afford the $15k+ FSD.

It’s not even level 3 yet and you’re talking about level 4?

 

[diplomat33]

It’s not even level 3 yet and you’re talking about level 4?

The SAE levels are not sequential. FSD Beta does not have to be L3 first. It is possible for FSD Beta to skip L3 and go straight to L4. In fact, there are plenty of AV systems (Cruise, Waymo etc) that started at L4 and were never L2 or L3.

 

[DanCar]

I agree. Keep in mind, If it ever goes level 4, it's going to be way more than $12k.

Better to maybe buy the dip right now and if it goes level 4, stock sky rockets and then you can afford the $15k+ FSD.

I expect competition with Mobileye around then in 4+ years. If one keeps their cars less than 3 years, then I would say that line of thinking isn't a good idea. Even if it is just level 3 I think the value proposition goes up substantially, up from current ~$3K price premium for a used FSD Tesla car.

 

[dbldwn02]

It’s not even level 3 yet and you’re talking about level 4?

Sorry, should have prefaced with... "I don't think my 2017 Model S will ever be a robo-taxi"

Thought that went without saying these days. LOL.

 

[redmunds]

The SAE levels are not sequential. FSD Beta does not have to be L3 first. It is possible for FSD Beta to skip L3 and go straight to L4.

At the slow incremental pace that the software is being developed, how could Tesla FSD possibly skip L3 and achieve L4? We're the beta testers. We're the ones they release it to. Who else are they going to get to secretly test the software before releasing it to us? I can't see that happening.

 

[JulienW]

At the slow incremental pace that the software is being developed, how could Tesla FSD possibly skip L3 and achieve L4? We're the beta testers. We're the ones they release it to. Who else are they going to get to secretly test the software before releasing it to us? I can't see that happening.

Tesla could offer L3 tomorrow (except for needing approval). They could easily offer a VERY limited L3 system similar to Mercedes' or even more advanced. All Tesla would have to do is narrowly define the ODD.

 

[dbldwn02]

Tesla could offer L3 tomorrow (except for needing approval). They could easily offer a VERY limited L3 system similar to Mercedes' or even more advanced. All Tesla would have to do is narrowly define the ODD.

Without getting too deep into the technical definitions of the SAE Levels (there's plenty of other threads to argue that mess), L3 is going to require Tesla assume some responsibility. I don't think they're willing to put the money where their mouth is, at least not for another several years.

 

[eevee-fan]

Without getting too deep into the technical definitions of the SAE Levels (there's plenty of other threads to argue that mess), L3 is going to require Tesla assume some responsibility. I don't think they're willing to put the money where their mouth is, at least not for another several years.

Well, Elon put his money where his mouth was and that did not work out too well. So, yes, I agree the board will be hesitant to approve L3.

 

[gsmith123]

At the slow incremental pace that the software is being developed, how could Tesla FSD possibly skip L3 and achieve L4? We're the beta testers. We're the ones they release it to. Who else are they going to get to secretly test the software before releasing it to us? I can't see that happening.

Lets say Tesla has an FSD build they think could release as an L3 system. There's not that much incentive for them to actually do that.

That would not improve the quality of testing they get from all of us. L2 is perfectly fine for us to keep testing and giving Tesla telemetry.

Pros:
- take rate and value of FSD increases

Cons:
- Direct financial cost of liability
- Reputational harm from crashes / problems
- problems require more urgent fixes for an L3 system
- it's harder and less safe to iterate on
- higher regulatory scrutiny

I could see Tesla foregoing an L3 system for a good long time and sticking with fully supervised L2 testing. At least for 2-3 more years, but potentially a lot lot longer.

They could even keep consumer FSD L2 while at the same time internally testing and deploying L4 geofenced versions (like Waymo, Cruise, etc).

 

[JulienW]

Without getting too deep into the technical definitions of the SAE Levels (there's plenty of other threads to argue that mess), L3 is going to require Tesla assume some responsibility. I don't think they're willing to put the money where their mouth is, at least not for another several years.

Not the point of my post (read the post I quoted). The point is Tesla (or any car company) doesn't have to offer L3 before it offers L4. Tesla doesn't HAVE to offer L3 first in order to develop L4 because they COULD offer L3 now (I understand all the SEA hoops but strictly referring to the ABILITIES of the system).

 

[dbldwn02]

Not the point of my post (read the post I quoted). The point is Tesla (or any car company) doesn't have to offer L3 before it offers L4. Tesla doesn't HAVE to offer L3 first in order to develop L4 because they COULD offer L3 now (I understand all the SEA hoops but strictly referring to the ABILITIES of the system).

Yea I wasn't disagreeing with you.

A lot of car companies have a level 2 system that can work as level 3, if they removed the nags and set usage limits. My Ioniq5 for example, does better auto steer (HDA2) and highway driving than my Tesla does. But they're also not advertising Full Self Driving...Tesla has been advertising Hands Off "Fool Self Driving Capability" for 8 years and it's still anything but that.

2016 demo video: "The car is driving itself. The driver is only there for legal purposes."

 

[redmunds]

Without getting too deep into the technical definitions of the SAE Levels (there's plenty of other threads to argue that mess), L3 is going to require Tesla assume some responsibility. I don't think they're willing to put the money where their mouth is, at least not for another several years.

Yes, but to be clear, Tesla will have to assume some responsibility for all SAE levels L3 and above.

 

[Daniel in SD]

That would not improve the quality of testing they get from all of us. L2 is perfectly fine for us to keep testing and giving Tesla telemetry.

It would make the testing so much more efficient. The vast majority of FSD disengagements are not necessary. An L3 system with no safety driver would only disengage on impact so it would let Tesla know exactly what the real bugs are.

 

[Daniel in SD]

All Tesla would have to do is narrowly define the ODD.

And enforce the ODD, the user is not responsible for knowing the ODD. This seems like it would be difficult. For example the Mercedes has microphones in the wheel wells to detect if the road surface is wet (and probably if it has sand on it or something else that could reduce traction).

 

[gsmith123]

It would make the testing so much more efficient. The vast majority of FSD disengagements are not necessary. An L3 system with no safety driver would only disengage on impact so it would let Tesla know exactly what the real bugs are.

That makes no sense.

If Tesla wanted to ignore driver disengagements they can easily do that.

Teslas whole strategy for FSD beta testing is to use humans monitoring an L2 system to figure out where it doesn’t work.

 

[Daniel in SD]

That makes no sense.

If Tesla wanted to ignore driver disengagements they can easily do that.

Teslas whole strategy for FSD beta testing is to use humans monitoring an L2 system to figure out where it doesn’t work.

I think people would not be happy if FSD did not allow disengagements! I also think that would not shield Tesla from liability.

 

[gsmith123]

That’s not what I said at all.

 

[AlanSubie4Life]

Teslas whole strategy for FSD beta testing is to use humans monitoring an L2 system to figure out where it doesn’t work.

How does this work?