Does everyone believe FSD will be viable in the next 3 years? 5 years? 10 years?

[Dids]

What's the point of robotaxis if there's a driver you have to pay to sit there. Are the customers considered the driver if they sit in the driver seat? How can the robotaxi go pick them up?

They don't. I was asking about regulations. Clearly regulations are different in Arizona.
https://www.google.com/amp/s/www.th...-driverless-cars-allow-more-customers-phoenix

Advertisement

 

[Crissa]

Someone has to be in the driver's seat at attention. Not drive it....

...That argument is really not going to serve you well when faced with liability or criminal law.

-Crissa

 

[Jhodgesatmb]

I'm curious why people think regulations have to "approve" FSD. Is FSD illegal? Wouldn't you first have to create regulations barring it to then be able to approve it? Maybe laws that state an operator has to be a licensed... And for robo-taxis car for hire regulations?

It’s true that no one has to approve FSD. It is a product and a set of features. But it isn’t autonomous driving either. That would fall under NHTSA regulations and public policy, neither of which seem ready for Tesla’s FSD yet (which is ok since it isn’t ready yet either). I noted with interest the other day when it was announced that a company has been given the green light for driverless taxis in San Francisco. Another reason for me to skip driving into the city.

 

[cybertruck808]

Waymo this month has gotten permission for public self-drive taxis in Phoenix. Not a pilot program on limited pre-selected customers and routes and safety drivers... But open to the public.

So we have within five years, bam.


It may be awhile until that's legal. You're still responsible for the car if something happens (say, someone hits your car, rather than the auto-driver screwing up) so that'll be a legal mess.

I've driven with assist features before, and it's like having a child helping you drive. They can shout if they see something wrong. Current Tesla is like driving with someone on a learner's permit. You gotta still be attentive, but they carry out most of the actions. That's still alot of work to do.

And if you want more than Autopilot, which is just fancy cruise-control with safety features, you need the FSD package. https://en.m.wikipedia.org/wiki/Tesla_Autopilot#Driving_features

Sensing lights and stop signs (which is available in some other cars) is a FSD feature, not an Autopilot one. And even at $8k I'm signing up for this. To always have another driver - even if they're not great - with you at all times is really awesome. Hey, FSD, pull the Cybertruck into that parking spot so I don't have to squeeze out of it. Thanks, FSD.

-Crissa

Good news! I am still skeptical of a nationwide adoption; I think we are likely to see a city by city approach in the interim for a few cities, but very skeptical of a nationwide adoption. That said, I hope I'm wrong. And even if I'm not wrong, I'm seriously rethinking canceling the FSD option due to the price continuing to increase by a fair amount. Yes, I can always add later, but these increases seem to be steady and high.

 

[cybertruck808]

Someone has to be in the driver's seat at attention. Not drive it....

It's a mixed bag. Some states have passed legislation explicitly authorizing autonomous driving; thus, autonomous driving is legal in those states. Other states have only created committees to study and it you still need to have someone driving the vehicle for driving to be legal. Here's an article from 2019 detailing some things: https://www.thedrive.com/tech/29338/what-are-the-regulatory-barriers-to-full-self-driving

Note, there are no national guidelines on this issue currently (or rather nothing binding), but that could easily change.

 

[Jhodgesatmb]

This topic keeps resurfacing and it is understandable because it is a lot of money to put up to possibly be someone's guinea pig. I was encouraged by someone's question during the Tesla quarterly report as to whether Tesla would consider having the FSD option go with the driver rather than the car and they said they would think about it. I was also encouraged when Tesla said they are going to implement waypoints on navigate on autopilot. That said, I think you should assume that the regulations and public policy about autonomous driving is out in the future somewhere (whether it is 5 or 10 years doesn't really matter), but the capability to drive autonomously is here today and you need to decide whether the capability is something you need and not be so concerned with the policy. If you cannot get your head around the distinction, then the amount of money cannot be justified. I know that it (the capability) will be much better by the time we see Cybertrucks in our driveways, and that I want the kind of assistance the capability will provide going forward, to secure greater autonomy, so I am in, but the decision is personal for everyone.

 

[T3slaDad]

All great comments so far! A big takeaway is that you need to look at not only how quickly the development of FSD is going (very fast, Tesla is putting billions into it), but also how your state is handling their reception to the idea of level 5 autonomy and vehicle depreciation (or appreciation).

FSD is developing very fast and frankly, having someone spend billions to sell me amazing driver-assist features for $7k (early reservation holder) even with today's standards is great! I have EAP on my M3 (AP/EAP/FSD stays with the car by the way, not the owner) and it's AMAZING!!! Sure, it's not perfect but I have taken 1500+mi trips with a whopping 1-2 interventions, and every update just gets better. Many trips I take may require 1 intervention, but for the most part it's practically flawless already. I almost pulled the trigger for FSD when it was a $3k upgrade and didn't, now regretting it because I love to be part of development and growth. Watching these FSD beta tester videos gives me total confidence in how fast the technology is growing: you really can't grasp how good it is until you regularly use EAP.

Anyway, where I live seems to be open to the idea of L5 autonomy. I wouldn't be surprised if within a year of FSD being complete, they give everyone the thumbs up to use it. That may or may not be the case where you live, you need to do your homework. If your CT is ready for amazing FSD, it could still be a long time if your legislation isn't open to the idea of enabling it (look at Europe for example, they have a dumbed down EAP package heavily due to legislation in many countries).

All that aside, let's talk about depreciation. Anyone with FSD on their car today is essentially going to have a theoretical $2k higher valuation on their vehicle come Monday. Yes, a car that's not 60yrs old that appreciated in value overnight! If you have the chance to lock in FSD on your CT for $7k, 8k, 10k, etc, as long as the cost for FSD on the new CT is more than what you're locked in for, your CT just became worth more. The sooner you lock in FSD, the more it's worth as we near CT launch and FSD price hikes. The price of FSD has gone up $3k this year alone, and $3k over the last 2 years (it fluctuated up and down a lot more back then, but is now stable in going only up). Assuming FSD is $15k by CT launch, your CT is already valued at $7-8k more than what you bought it for - a 17% increase in value on day 1!

In the end, it's not mine or anyone else's job to convince you what to do, but I hope this sparked a few ideas to take into consideration on if FSD is worth it to you. There a lot more at play than simply saying "is FSD feature complete yet?" like so many users are asking.

 

[Newton]

I really like the idea of FSD staying with the driver. Since it seems all their cars will all have the systems Installed for fsd.. why not just have it available for use by everyone, rather then just the cars that have it unlocked.
Therefore if you have the fsd monthly subscription you can use it in any tesla u get in, or maybe a founder's/ early adopters lifetime reward for those who are willing to pay the 10k... Because 10k or more extra, everytime you want to upgrade your vehicle is just Ridiculous in my opinion.

 

[Crissa]

I noted with interest the other day when it was announced that a company has been given the green light for driverless taxis in San Francisco. Another reason for me to skip driving into the city.

...They've been driving in the city for almost ten years. You hadn't noticed? They're the best drivers on the road.

-Crissa

 

[Iacemoe]

It all comes down to the data. Once Tesla can show that it has safety data that far exceeds human capability while fully autonomous without intervention, its only a matter of time. Whamo already is getting the ball rolling regulation wise in geo fenced areas, so Tesla isn't a first mover there. Should help to speed up regulation issues, or at least get regulators thinking about the topic

I think it will be amazing to see how fast this neural net develops. My bet is the pace of improvement is going to surprise most.

 

[FutureBoy]

It all comes down to the data. Once Tesla can show that it has safety data that far exceeds human capability while fully autonomous without intervention, its only a matter of time. Whamo already is getting the ball rolling regulation wise in geo fenced areas, so Tesla isn't a first mover there. Should help to speed up regulation issues, or at least get regulators thinking about the topic

I think it will be amazing to see how fast this neural net develops. My bet is the pace of improvement is going to surprise most.

TLDR: FSD rollout is going to look somewhat slow at first but relatively soon is going to start going very fast and surprise many drivers. The main potential issue will be in balancing skillful and conservative drivers with the random chaos of life.

———————

The skill of the neural net will be dependent on 2 things, the number of example cases it injests, and the number of new scenarios that are included in those cases.

At the beginning, FSD is going to have a lot of new scenarios per number of injests. But the number of injests will be small because the current number of vehicles using FSD beta version is still small.

As FSD gets more skillful, FSD will get rolled out to more vehicles. With more vehicles, the number of injests will go up. And the number of new scenarios will go up too. At some point though the number of new scenarios seen will max out because most vehicles will be driving the same routes under similar conditions as were already collected.

From there, the race will be to get as many vehicles out there collecting new scenarios as possible. And eventually the number of injests will be huge but the number of new scenarios will drop to a level of diminishing returns.

The whole way through this process Tesla, regulators, drivers, and the general public would LLC be weighing in on their views of how well FSD is doing and letting us know if they trust FSD yet or not. Each group will also have their own incentives for either trusting or doubting the skill of FSD.

All this to say, in the beginning I believe that Tesla will be extra cautious rolling out FSD. But that extra caution will still look fast to the uninformed person. But once we hit a turning point of acceptance for FSD, there is going to be exponential growth in the skill level of FSD. And barring any especially shocking negative effects (like FSD plowing through a group of toddlers on a walking field trip) there will be a really quick acceptance and rook out of FSD.

The trick will come in Tesla balancing the tool out to increase the number of injests while keeping the number of new scenarios to a manageable level so that they can reduce significantly the number of negative incidents. But because that is dependent on the trustworthiness of the drivers, that balance will be difficult to achieve.


TLDR: FSD rollout is going to look somewhat slow at first but relatively soon is going to start going very fast and surprise many drivers. The main potential issue will be in balancing skillful and conservative drivers with the random chaos of life.

 

[Jhodgesatmb]

...They've been driving in the city for almost ten years. You hadn't noticed? They're the best drivers on the road.

-Crissa

I hadn’t noticed because I avoid the city. I hate cities, not because of driverless cars. When all the cars are driverless, and the bikers become conscientious, and the pedestrians act like human beings, and the prices become reasonable, then maybe I’ll feel more comfortable, but they will still be packed with people and I have more of a rural mentality.

 

[Jhodgesatmb]

It all comes down to the data. Once Tesla can show that it has safety data that far exceeds human capability while fully autonomous without intervention, its only a matter of time. Whamo already is getting the ball rolling regulation wise in geo fenced areas, so Tesla isn't a first mover there. Should help to speed up regulation issues, or at least get regulators thinking about the topic

I think it will be amazing to see how fast this neural net develops. My bet is the pace of improvement is going to surprise most.

I haven’t seen any footage of FSD in downtown [anywhere, let alone San Francisco] areas. Has anyone? Maybe Tesla is requiring that early beta testers only use it at night or in light traffic so far.

 

[ajdelange]

It all comes down to the data.

Alas, that's not the case at all. The fact that a FSD system can provide two more nines than a human drier will be little comfort to the family of the guy that got killed, or the lawyers his family retains, by the FSD system. A drunk running over a school child is hardly news any more but a Tesla on autopilot killing someone really is.

One thing that never gets talked about here is the ethics problem. Given that FSD has to kill its passngers, 3 prominent heart surgeons on the way to save the life of a little girl, or 5 illegal immigrants on the side of the road which should it do? Get a bunch of your liberal and conservative friends together for a "friendly" debate on that one. Then ask yourself how long it will take the regulatory agencies to come up with guidance on such questions.

 

[Jhodgesatmb]

Alas, that's not the case at all. The fact that a FSD system can provide two more nines than a human drier will be little comfort to the family of the guy that got killed, or the lawyers his family retains, by the FSD system. A drunk running over a school child is hardly news any more but a Tesla on autopilot killing someone really is.

One thing that never gets talked about here is the ethics problem. Given that FSD has to kill its passngers, 3 prominent heart surgeons on the way to save the life of a little girl, or 5 illegal immigrants on the side of the road which should it do? Get a bunch of your liberal and conservative friends together for a "friendly" debate on that one. Then ask yourself how long it will take the regulatory agencies to come up with guidance on such questions.

I don’t like what you are saying but I do respect them and believe them. Sadly, the SAE defined functional capabilities for levels of autonomy but they could not and would not be able to address psychological, emotional, ethical, or moral issues therein, and society may only be able to look at it that way.