I just can’t understand why you’d give up control to a system that’s clearly still in beta. The cost of failure is way too high to take the risk.

Because they're led to believe all this public beta "testing" actually does something to improve the system. So they go out on roads, enable FSD and click the feedback button when it makes a mistake. It's pure theatre as it does nothing to a system fundamentally poor.

From my observations I can’t show causality, but the three situations where I used the feedback button the most all got addressed.

- my city town inexplicably decided to make dashed white lines on the sides of bicycle lane sharing markers. The Tesla would see dashed white lines, which mean separate lanes going the same way and dance around trying to get into a lane but they were all too narrow. The traffic lane was double wide to account for church street parking, but not marked as such. So it the markings made a tiny center lane and two side lanes that were just a bit too small.

- on winding tree lined roads, when an on coming car first appeared around a bend it would assume the car might be going straight and would hit us. There would be a quick deceleration and dodge toward the shoulder until it realize the oncoming car was actually curving and would probably stay in its lane.

- when the pavement expands to include an unoccupied, unmarked parking lane for a short stretch the car would assume that the road suddenly got very wide lanes and it was too far to the center and jerk over.

At this point FSD does a passable job on my regular daily drives, except one intersection where something in its map is wrong, it wants to exit the right turn lane just before turning right. I don’t let it make turns in contested intersections yet, because it is a little timid and might confuse drivers. But that is improving. Just not there yet.

For exceptions, it does well finding and dealing with construction areas, garbage trucks, bicycles, and pedestrians.

I’d appreciate if they’d add “slightly dodge roadkill”. It will merrily run over the same dead squirrel four times a day with precision.

They /are/ the training data and trainers in a very literal sense. Every time they correct or take control over FSD I imagine it gets fed back into the training data. Will it ever create a fully self driving system? Maybe.

That's not true and Karpathy has confirmed it multiple times. They can create a fingerprint for data they want the system to feed back to the mothership and they may manually label datasets fed back through the report button on the MCU, but disengagements are not a signal they use at all.

It's useful for polishing an already mature system. But FSD is poor at even doing the basics because of their (lack of) sensors. They need more than low quality 30 second video clips if they ever want a usable autonomous system.

I’m sure other people use FSD as a bad chauffeur, but around town I use it when I’m willing to be extra alert and train it. There is a button I press when the car has messed up and the surrounding data gets sent back to the Tesla mothership as a defect.

If I’m not willing to be in vigilance mode I just drive myself.

thanks for sharing your individual lack of experience with a broadly systemic issue (overconfidence in FSD, believing their marketing, etc)

And the beauty of not advertising that if an accident is coming the system deactivates itself to avoid any legal responsibilities. Autopilot that deactivates itself. Nice

Your "straw man!" claims ignore that for every 'safety' related comment or action Tesla has done, they've had to be dragged kicking and screaming into it.

First there was zero monitoring of attention. No steering wheel torque, nothing.

Then when pushed and threatened they added it, and set the "hands on steering wheel" interval to "ONCE every fifteen MINUTES".

Then they had to be threatened to reduce that to a sensible value.

All the while, Tesla: "the driver is only in the seat for legal purpose - the car is driving itself", or "use Summon to bring your car to you while you deal with a fussy child".

So you can call them lazy edge-case idiots all you like.

But you can't pretend that Tesla hasn't absolutely fostered and nurtured this culture, because they have.

I covered all that.

Someone abandons common sense and abdicates responsibility for themselves in favor of believing the hype from a money-crazed billionaire?

Prime candidates for Darwin awards, whether they own a tesla or not.

We are terrible at maintaining 100% concentration over extended periods of time while nothing is happening. People fall asleep driving. The mind starts to wander. We get distracted by our children.

I don't think it's about believing the hype. It's about having a system put in place which encourages you to stop paying attention.

I do think there have been great improvements in car safety thanks to AI. My car self brakes if it detects a pedestrian while I'm in reverse. It flashes lights in my side mirrors if another vehicle passes by. I think those systems are safety first and if you buy a car they are the responsibility of the car maker and not the buyer. I don't think you would expect people checking the airplane model when taking a flight. It's the responsibility of the airline and the manufacturer.

"People who believe the system they paid 10-12k for called 'full self driving' should be able to drive itself are idiots" is a pretty special class of Elon bootlicking.

People who understand the system don't buy it because they realize it's a scam, everyone who bought it was duped. The most valuable thing at Tesla is the list of rubes they have.

Many of whom are well equipped to know better.

> You are not (currently) using an autonomous car. If you think so you are an idiot.

Let's define autonomous as self driving. According to their marketing page[1]:

> Tesla cars come standard with advanced hardware capable of providing Autopilot features, and full self-driving capabilities—through software updates designed to improve functionality over time.

So it has capabilities of autonomous driving, which means a Tesla can be considered an autonomous car. Unless I'm an idiot for trusting what their marketing page tells me, of course.

1: https://www.tesla.com/autopilot

They phrase it weirdly (accidentally/possibly/probably/certainly (take your pick) in order to confuse the readers), but all that says is that they eventually will provide the software running on the hardware you buy today to actually do it.

I disagree. When I read this:

> Tesla cars come standard with advanced hardware capable of providing Autopilot features, and full self-driving capabilities—through software updates designed to improve functionality over time.

What I understand is this:

Tesla cars come with hardware capable of providing features for the Autopilot system and come with full self driving capabilities. The functionality related to these capabilities will be improved over time through software updates.

If they meant that there will eventually be software that works with the hardware in the system, but there isn't software today, they should word it differently.

It is intended to be parsed as:

The cars come with advanced hardware. What is the hardware capable of? It is capable of self driving. How will the hardware achieve the thing it is capable of? Through software updates (that have not yet landed)

That might be what some weaselly lawyer argues it means in court, or what someone deeply familiar with it all would infer. But in no way is that how any reasonable lay person would read it.

"Improve functionality over time" implies that the basic functionality of full self-driving is there already, and that it will be improved. No reasonable person would take it to mean "does not exist yet, and may never do".

They would also take "capable of" to mean "right now", not at some nebulous point in the future, especially given the context of the feature being a paid upgrade.

The definition of "full" is "not lacking or omitting anything; complete". One might reasonably expect "full self-driving" to mean a human does not need to intervene or monitor. I.e. What the industry commonly calls autonomous driving level 5. They clarify what they mean elsewhere on their web site:

> Autopilot, Enhanced Autopilot and Full Self-Driving Capability > are intended for use with a fully attentive driver, who has their > hands on the wheel and is prepared to take over at any moment. > While these features are designed to become more capable over time, > the currently enabled features do not make the vehicle autonomous.

So they don't even claim level 1 autonomy.

Even with your version, the hardware may very well never be capable of full self-driving at all. There's no evidence to suggest it will be - hopes, dreams and promises don't count. And they've done an upgrade on the earlier-gen Model S cars already, where they presumably made similar claims.

The whole thing is misleading at best, downright trading standards fraud at worst.

For what it's worth, my Model 3's carriageway departure warning kicks in randomly and the amount of phantom braking it does even on basic cruise control is terrible. Autopilot is worse. I find monitoring it in autopilot mode significantly more stressful than just driving the car myself. I wouldn't trust FSD at all, and certainly not enough to want to fork out the silly money it is.

"The driver is only in the seat for legal purposes - the car is driving itself".

> but all that says is that they eventually will provide the software running on the hardware you buy today to actually do it

Most of their language about "eventually" very very strongly implies that the only hold up is regulatory, not software.

>Unless I'm an idiot for trusting what their marketing page tells me, of course.

Oh, you sweet summer child...

> the most advanced ML learning system ever

Imagine believing this.

I can believe this but also believe that our "most advanced ML systems" aren't up to the autonomous driving mark yet.

You would have to believe that Tesla, particularly, had that system, and not Google or DoD.

Tesla has something much more valuable: pigeons and dupes.

That is terrifying! I try to stay away from Teslas I see on the road because they are completely unpredictable. I think I'd prefer being near a poor driver with their attention on the controls to Autopilot with a poor driver not paying attention.

Any test on public roads of a system that performs like this should have a professional team actively, directly monitoring it. Not just any rando who happened to be able to afford a Tesla.

The number of times I see Teslas suddenly swerve partially into the next lane or stomp the brakes for no reason on the freeway with nobody around them is ridiculous.

To top things off, one time I saw this happening on 280 in the Bay Area and I decided for safety's sake to pass them. I looked inside and it was a baby in a car seat in the back and a guy at the driver seat with both hands off the wheel holding onto a phone that he was absorbed by.

there's a lot more in a youtube channel that was posted here a while back, the entire channel was dedicated to catching autopilot's mistakes.

edit: whoa somebody is flagging all of critical comments of tesla!

Meanwhile, I've been driving mine daily with autopilot for over a year daily, and can count on one hand the amount of times I've needed to take manual control.

My anecdotal experience is they're incredibly predictable and with an attentive driver (as you should always be in any car), completely a non issue.

Musk, every year for the last seven years now:

"Full Self Driving is coming, this year."

Not even close.

This is how he holds at bay the two factions of lawyers:

1) Those ready to file a misrepresentation or breach of warranty contract claims;

2) Those ready to file a products liability, wrongful death or failure to warn tort lawsuit.

No. It is how he suckers people into paying for it now.

Fending off lawyers is based on the fact it it was only Musk who said it, courts recognizing that only an idiot would believe him.

The best part is people paid $12000 for the compute "hardware."

Yet, all the cars have it. So they are also paying for all the other cars' hardware.

Jeez.. everyone's a critic... Tesla's got this! /s

The most sad and ironic about your comment, is that had you not added the /s tag, a few on this thread would assume you were serious :-)

Indeed. /$

This is why I tell my wife to move over when you see a Tesla.

The issue with these AI systems is that the corner cases in the real-world can be fatal.

Humans are pretty good, on the whole, at dealing with corner cases.

We're just not great at reaction time and constant vigilance

On the other hand, each human needs to individually figure out (or be taught) how to deal with each corner case.

When an AI system learns it, the company can "teach" the entire fleet all at once.

You can't make an apples-to-apples comparison between humans and the current generation of "AI".

A human does not need to be taught how to handle every edge case because we have the ability and intuition to think on the spot and create solutions on the fly. It's incredible how fast our brain can take in information and formulate a plan.

Meanwhile, if an "AI" has not specifically been trained against a certain variable or situation, it will have no idea how to handle it. It can't actually create a solution like a real human brain can.

And yet, all these individual Teslas make the same basic mistakes. What is it: lowest common denominator?