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

FSD v12.x (end to end AI)

This site may earn commission on affiliate links.
#11,782
Pdubs said:
I can’t believe alansubpar actually downvoted an experience post. /s (I totally believe it)

I find your level of retention amazing. Do you drive with a recorder and take notes? I can’t remember the last 10 miles I drove.
Click to expand...
No recorder, no notes.

As it happens, I haven't been driving a ton with 12.3.4 and FSDS. Or, on this eclipse trip I took earlier this month, it was on the SO's Model Y and it was +95% on highways.. so not a really good test of streets and city driving. Even if it was a 3000 mile trip. And it was 12.3.3, anyway.

So, on the trip to South Mountain and back, knowing that it wasn't going to be that easy (we're talking Jersey, during the week, including rush hour, and some pretty blame congested places), I tried to pay serious attention to what the car was doing. Remember, I'm one of the people who were running the 10.x variants a couple of years ago; and I know just how dangerous it can be to just let the car do its own thing. What's that phrase: "It's amazing how the prospect of immanent death sharpens the mind."

Encountering fun stuff like that yellow gear and random contractors walking back and forth across the street and letting the car stop for them, with my foot hovering over the brake pedal and a little adrenaline running helps one remember as well.

Heh. So, the SO's MY has had EAP on it since shortly after getting the vehicle. So, when the price of FSDS dropped from 12K to 8K, the price for us to upgrade from EAP dropped from $6k to $2k. Yeah, yeah: At the $99/month rental price, it's going to be 20 months before buying FSDS for the car is more cost efficient than buying it outright. But $2k is low enough to tempt the eye, so, over several days, we thought about it. I will say that neither of us had reached a solid decision whether to do it or not, but we had a serious debate yesterday, back and forth, and decided to go ahead.

First off: Listened to the earnings call a couple days ago. Didn't precisely take notes, but may as well have. And looked at the stockholder pack as well. Tesla is currently testing candidates for Robotaxi. There are bugs, but the people from the company at the earnings call are convinced that said bugs are fixable. That's a new statement. Further, the shareholder deck included some pictures of the Robotaxi app. There was also some discussion that the app would allow different levels of Who Gets To Ride In The Robotaxi: Near family, family and friends, the general public, and so on. Musk reiterated that, once RT is up and running, Tesla would be running a fleet of these things, never mind people who've bought the cars. Finally, there was Musk, strongly requesting that investors (and I think we're talking mainly the analysts) to get into a Tesla with FSDS and Take A Drive. Right now.

All of this indicates to me that RT, and by extension, FSDS is about to go seriously mainstream. Yeah, that negates all the naysayers around here who keep on saying anything from "Never" to "Five years from now." But it sure didn't sound aspirational. It sounded like product development.

What this means: Tesla's not shy about shifting prices on things up and down, either to get more product out the door, or handle increased COGS, or to push some behavior they think is appropriate. So.. despite the current lowering of FSDS pricing, I strongly suspect it won't stay that way.

Second: The main reason the SO got EAP after living with Basic Autopilot for a time was EAPs safety features. Want to shift lanes on an interstate? Having EAP means not having to worry about one's blind spots. Either for another car hovering in one or zooming up from the rear or side - the car will see it not move to end up in an accident. What with FSDS's improvements.. it might not be Quite There yet with all those 9's people chatter on about, but if RT is coming, it will be. Even a minor accident is going to cost a heck of a lot more than $2k.

The one solid reason the SO came up with for not getting FSDS was an interesting one. She happens to be a Human Factors engineer and has the degrees to prove it. A good portion of HF is Preventing Accidents. At the minimum, we're talking about making sure the dials and controls on aircraft are standardized, so a pilot stuck in an unfamiliar aircraft doesn't auger said aircraft into the ground at the first opportunity. Or that the command discipline in an aircraft cockpit, or a ship, or in an operating room doesn't lead to bad outcomes.

On this last long trip we took, each of us managed to violate the 85 mph limit for Autopilot and had to drive some 30 miles or so sans any of that automation to our next scheduled stops. And both of us.. well, we did it, without scaring anybody, and it's not like we don't have decades of practice driving cars without any kind of cruise control. But, over the past six years we've just gotten used to driving with the technology. And found it disconcerting and a little awkward to drive at 80 or so, manually. It's kind of like airliner pilots: They spend nearly all of their time on autopilot, so when it comes time in an emergency to Just Fly The Plane, the reflexes that would normally be present aren't all what they're supposed to be. Especially if one is in Low and Slow, or (paradoxically) High and Fast, at the boundaries that those big guys can fly at. Which is why airline pilots for reputable airlines spend a significant amount of time in simulators.

With FSDS, the problem is that, if it gets good enough, those cutting edge reflexes we all have when driving local streets might get blunted. Which was her point.

So, we both swore that we'd Very Definitely Keep In practice. Discussed the low price some more, then pulled the trigger. Now we've got two cars on paid-for FSDS.
 
  • Like
  • Funny
  • Informative
Reactions: Z_Lynx, tmoz, Electroman and 7 others
#11,783
Tronguy said:
First off: Listened to the earnings call a couple days ago. Didn't precisely take notes, but may as well have. And looked at the stockholder pack as well. Tesla is currently testing candidates for Robotaxi. There are bugs, but the people from the company at the earnings call are convinced that said bugs are fixable. That's a new statement.
Click to expand...

New?

Elon Musk in 2019 said:
I feel very confident predicting that there will be autonomous robotaxis from Tesla next year
Click to expand...

Elon Musk in 2019 said:
Next year for sure, we will have over a million robotaxis on the road. The fleet wakes up with an over-the-air update. That's all it takes.
Click to expand...
 
  • Like
Reactions: mtndrew1, Ben W, Matias and 1 other person
#11,784
Tronguy said:
But it sure didn't sound aspirational. It sounded like product development.
Click to expand...
LOL. Watch the 2019 Autonomy Day presentation. That’s just the way Elon talks.
They even had a robotaxi app back then!
1714068329901.png
 
  • Like
  • Informative
Reactions: tmoz, AlanSubie4Life, mtndrew1 and 6 others
#11,787
Knightshade said:
Because the default behavior of the system is...not to speed. It sets speed max to observed (or mapped) speed limit.

The only way it can ever speed is if the human instructs it to (either via settings, scroll wheel, accelerator, etc).

Whereas the DEFAULT prior to the recall was to make illegal stops.


That said- I assure you any driverless system will be required not to speed either.

And AFAIK only place I can find "safety" referenced is Teslas description of the defect in their voluntary recall notice so apparently Tesla thinks it's a safety issue.
Click to expand...
Yes, Tesla should have made California stops an option with default being full stops.
 
  • Like
Reactions: dhanson865, pilotSteve, FSDtester#1 and 1 other person
#11,788
Tronguy said:
No recorder, no notes.

As it happens, I haven't been driving a ton with 12.3.4 and FSDS. Or, on this eclipse trip I took earlier this month, it was on the SO's Model Y and it was +95% on highways.. so not a really good test of streets and city driving. Even if it was a 3000 mile trip. And it was 12.3.3, anyway.

So, on the trip to South Mountain and back, knowing that it wasn't going to be that easy (we're talking Jersey, during the week, including rush hour, and some pretty blame congested places), I tried to pay serious attention to what the car was doing. Remember, I'm one of the people who were running the 10.x variants a couple of years ago; and I know just how dangerous it can be to just let the car do its own thing. What's that phrase: "It's amazing how the prospect of immanent death sharpens the mind."

Encountering fun stuff like that yellow gear and random contractors walking back and forth across the street and letting the car stop for them, with my foot hovering over the brake pedal and a little adrenaline running helps one remember as well.

Heh. So, the SO's MY has had EAP on it since shortly after getting the vehicle. So, when the price of FSDS dropped from 12K to 8K, the price for us to upgrade from EAP dropped from $6k to $2k. Yeah, yeah: At the $99/month rental price, it's going to be 20 months before buying FSDS for the car is more cost efficient than buying it outright. But $2k is low enough to tempt the eye, so, over several days, we thought about it. I will say that neither of us had reached a solid decision whether to do it or not, but we had a serious debate yesterday, back and forth, and decided to go ahead.

First off: Listened to the earnings call a couple days ago. Didn't precisely take notes, but may as well have. And looked at the stockholder pack as well. Tesla is currently testing candidates for Robotaxi. There are bugs, but the people from the company at the earnings call are convinced that said bugs are fixable. That's a new statement. Further, the shareholder deck included some pictures of the Robotaxi app. There was also some discussion that the app would allow different levels of Who Gets To Ride In The Robotaxi: Near family, family and friends, the general public, and so on. Musk reiterated that, once RT is up and running, Tesla would be running a fleet of these things, never mind people who've bought the cars. Finally, there was Musk, strongly requesting that investors (and I think we're talking mainly the analysts) to get into a Tesla with FSDS and Take A Drive. Right now.

All of this indicates to me that RT, and by extension, FSDS is about to go seriously mainstream. Yeah, that negates all the naysayers around here who keep on saying anything from "Never" to "Five years from now." But it sure didn't sound aspirational. It sounded like product development.

What this means: Tesla's not shy about shifting prices on things up and down, either to get more product out the door, or handle increased COGS, or to push some behavior they think is appropriate. So.. despite the current lowering of FSDS pricing, I strongly suspect it won't stay that way.

Second: The main reason the SO got EAP after living with Basic Autopilot for a time was EAPs safety features. Want to shift lanes on an interstate? Having EAP means not having to worry about one's blind spots. Either for another car hovering in one or zooming up from the rear or side - the car will see it not move to end up in an accident. What with FSDS's improvements.. it might not be Quite There yet with all those 9's people chatter on about, but if RT is coming, it will be. Even a minor accident is going to cost a heck of a lot more than $2k.

The one solid reason the SO came up with for not getting FSDS was an interesting one. She happens to be a Human Factors engineer and has the degrees to prove it. A good portion of HF is Preventing Accidents. At the minimum, we're talking about making sure the dials and controls on aircraft are standardized, so a pilot stuck in an unfamiliar aircraft doesn't auger said aircraft into the ground at the first opportunity. Or that the command discipline in an aircraft cockpit, or a ship, or in an operating room doesn't lead to bad outcomes.

On this last long trip we took, each of us managed to violate the 85 mph limit for Autopilot and had to drive some 30 miles or so sans any of that automation to our next scheduled stops. And both of us.. well, we did it, without scaring anybody, and it's not like we don't have decades of practice driving cars without any kind of cruise control. But, over the past six years we've just gotten used to driving with the technology. And found it disconcerting and a little awkward to drive at 80 or so, manually. It's kind of like airliner pilots: They spend nearly all of their time on autopilot, so when it comes time in an emergency to Just Fly The Plane, the reflexes that would normally be present aren't all what they're supposed to be. Especially if one is in Low and Slow, or (paradoxically) High and Fast, at the boundaries that those big guys can fly at. Which is why airline pilots for reputable airlines spend a significant amount of time in simulators.

With FSDS, the problem is that, if it gets good enough, those cutting edge reflexes we all have when driving local streets might get blunted. Which was her point.

So, we both swore that we'd Very Definitely Keep In practice. Discussed the low price some more, then pulled the trigger. Now we've got two cars on paid-for FSDS.
Click to expand...
I share your sentiments. But all I got from it was they will be "purpose built". I heard that as, anybody that bought one based on his earlier promises they would have a robotaxi were quite mistaken. I'd never put my current car out there and didn't get it with that in mind. (I just liked the tech and wanted to support it), but I didn't go in with blinders on, which I feel like a lot of posters did. As a shareholder, I was encouraged by the call and went deeper when $TSLA dropped, almost down to where I first went in.
 
  • Love
  • Like
Reactions: FSDtester#1 and zoomer0056
#11,790
Tronguy said:
No recorder, no notes.

As it happens, I haven't been driving a ton with 12.3.4 and FSDS. Or, on this eclipse trip I took earlier this month, it was on the SO's Model Y and it was +95% on highways.. so not a really good test of streets and city driving. Even if it was a 3000 mile trip. And it was 12.3.3, anyway.

So, on the trip to South Mountain and back, knowing that it wasn't going to be that easy (we're talking Jersey, during the week, including rush hour, and some pretty blame congested places), I tried to pay serious attention to what the car was doing. Remember, I'm one of the people who were running the 10.x variants a couple of years ago; and I know just how dangerous it can be to just let the car do its own thing. What's that phrase: "It's amazing how the prospect of immanent death sharpens the mind."

Encountering fun stuff like that yellow gear and random contractors walking back and forth across the street and letting the car stop for them, with my foot hovering over the brake pedal and a little adrenaline running helps one remember as well.

Heh. So, the SO's MY has had EAP on it since shortly after getting the vehicle. So, when the price of FSDS dropped from 12K to 8K, the price for us to upgrade from EAP dropped from $6k to $2k. Yeah, yeah: At the $99/month rental price, it's going to be 20 months before buying FSDS for the car is more cost efficient than buying it outright. But $2k is low enough to tempt the eye, so, over several days, we thought about it. I will say that neither of us had reached a solid decision whether to do it or not, but we had a serious debate yesterday, back and forth, and decided to go ahead.

First off: Listened to the earnings call a couple days ago. Didn't precisely take notes, but may as well have. And looked at the stockholder pack as well. Tesla is currently testing candidates for Robotaxi. There are bugs, but the people from the company at the earnings call are convinced that said bugs are fixable. That's a new statement. Further, the shareholder deck included some pictures of the Robotaxi app. There was also some discussion that the app would allow different levels of Who Gets To Ride In The Robotaxi: Near family, family and friends, the general public, and so on. Musk reiterated that, once RT is up and running, Tesla would be running a fleet of these things, never mind people who've bought the cars. Finally, there was Musk, strongly requesting that investors (and I think we're talking mainly the analysts) to get into a Tesla with FSDS and Take A Drive. Right now.

All of this indicates to me that RT, and by extension, FSDS is about to go seriously mainstream. Yeah, that negates all the naysayers around here who keep on saying anything from "Never" to "Five years from now." But it sure didn't sound aspirational. It sounded like product development.

What this means: Tesla's not shy about shifting prices on things up and down, either to get more product out the door, or handle increased COGS, or to push some behavior they think is appropriate. So.. despite the current lowering of FSDS pricing, I strongly suspect it won't stay that way.

Second: The main reason the SO got EAP after living with Basic Autopilot for a time was EAPs safety features. Want to shift lanes on an interstate? Having EAP means not having to worry about one's blind spots. Either for another car hovering in one or zooming up from the rear or side - the car will see it not move to end up in an accident. What with FSDS's improvements.. it might not be Quite There yet with all those 9's people chatter on about, but if RT is coming, it will be. Even a minor accident is going to cost a heck of a lot more than $2k.

The one solid reason the SO came up with for not getting FSDS was an interesting one. She happens to be a Human Factors engineer and has the degrees to prove it. A good portion of HF is Preventing Accidents. At the minimum, we're talking about making sure the dials and controls on aircraft are standardized, so a pilot stuck in an unfamiliar aircraft doesn't auger said aircraft into the ground at the first opportunity. Or that the command discipline in an aircraft cockpit, or a ship, or in an operating room doesn't lead to bad outcomes.

On this last long trip we took, each of us managed to violate the 85 mph limit for Autopilot and had to drive some 30 miles or so sans any of that automation to our next scheduled stops. And both of us.. well, we did it, without scaring anybody, and it's not like we don't have decades of practice driving cars without any kind of cruise control. But, over the past six years we've just gotten used to driving with the technology. And found it disconcerting and a little awkward to drive at 80 or so, manually. It's kind of like airliner pilots: They spend nearly all of their time on autopilot, so when it comes time in an emergency to Just Fly The Plane, the reflexes that would normally be present aren't all what they're supposed to be. Especially if one is in Low and Slow, or (paradoxically) High and Fast, at the boundaries that those big guys can fly at. Which is why airline pilots for reputable airlines spend a significant amount of time in simulators.

With FSDS, the problem is that, if it gets good enough, those cutting edge reflexes we all have when driving local streets might get blunted. Which was her point.

So, we both swore that we'd Very Definitely Keep In practice. Discussed the low price some more, then pulled the trigger. Now we've got two cars on paid-for FSDS.
Click to expand...

As you were paying very close attention to the call and slide deck, has the company disclosed any meaningful statistics on FSD? Disengagements per mile? Interventions per mile? Surely they know their target numbers as they've mentioned numerous times "ten times safer than a human."

1714076111533.png

1714076129642.png


Tesla shared the second graph on hitting 1bn miles on April 3rd. The "free trial FSD" started March 25th. The first graph (from the deck) is quite vague - the labels on the X axis are horrendous. Is that March 1st or March 31st?

You will also note the dates on the second graph are not spaced evenly. This is an attempt to make things look like they're growing exponentially.

You can just take the graph from 4Q23 and fill in Q1....
1714076625688.png
 
  • Disagree
  • Like
  • Informative
Reactions: winfield100, AlanSubie4Life, mtndrew1 and 1 other person
#11,791
Tronguy said:
All of this indicates to me that RT, and by extension, FSDS is about to go seriously mainstream. Yeah, that negates all the naysayers around here who keep on saying anything from "Never" to "Five years from now." But it sure didn't sound aspirational. It sounded like product development.
Click to expand...
No it is not.
Tronguy said:
So.. despite the current lowering of FSDS pricing, I strongly suspect it won't stay that way.
Click to expand...
Price of FSDS is going to collapse now that it is not FSD. Remember FSD is what gets more expensive over time.
EVNow said:
Tesla should have made California stops an option with default being full stops.
Click to expand...
Would still be broken.
DarkandStormy said:
This is an attempt to make things look like they're growing exponentially.
Click to expand...
Also note these are primarily not v12 miles as we move forward from March (the miles indicated as v12 are primarily v11 in the v12 release - otherwise the numbers don’t make sense).
 
Last edited:
  • Like
  • Disagree
Reactions: FSDtester#1 and Sharps97
#11,792
  • Informative
Reactions: DarkandStormy, JB47394 and AlanSubie4Life
#11,795
DarkandStormy said:
You will also note the dates on the second graph are not spaced evenly. This is an attempt to make things look like they're growing exponentially.
Click to expand...
I was about to say that that's fraud, but then I noticed that the whole graph is screwy, so I just think someone incompetent created it. The yellow numbers are the number of months between each x-axis label. I can't figure out how someone could screw up that much—am I missing something?

Tesla Graph Fail.jpg
 
  • Helpful
Reactions: JB47394
#11,796
PianoAl said:
I can't figure out how someone could screw up that much—am I missing something?
Click to expand...

My best guess would be:

1. They didn't log how many miles were driven on a regular basis. It was collected on an ad hoc basis, like upon certain version releases.
2. They treated the dates of those data points as an ordinal variable instead of continuous.
 
  • Helpful
Reactions: JB47394
#11,797
Tronguy said:
The main reason the SO got EAP after living with Basic Autopilot for a time was EAPs safety features. Want to shift lanes on an interstate? Having EAP means not having to worry about one's blind spots. Either for another car hovering in one or zooming up from the rear or side - the car will see it not move to end up in an accident. What with FSDS's improvements.. it might not be Quite There yet with all those 9's people chatter on about, but if RT is coming, it will be. Even a minor accident is going to cost a heck of a lot more than $2k.
Click to expand...

This.

I hate to admit it, but the main reason I subscribe to FSD is so I can tell the car to switch lanes on the Interstate and not worry about the blind spots. That one thing removes a lot of stress for me.

I make a lot of road trips and I'm not as young as I used to be. Actually, I guess no one is ....
 
  • Like
Reactions: tmoz, old pilot and Dennisis
You must log in or register to reply here.

Similar threads

F
Using FSD for 3 months - Is my experience typical?
Replies
69
Views
6K
AI, Autopilot, & Autonomous/FSD
UncleCreepy
UncleCreepy
G
  • Article
Tesla FSD - A Five Year Perspective
Replies
80
Views
11K
AI, Autopilot, & Autonomous/FSD
tevman
tevman
Top
Back
Blog
New
Forum list
Marketplace
Menu