FSD is not good enough for me. Sorry

[CT_AZ_4x4]

The level of defensiveness from some here is interesting. Hey, I respect your feedback and don't feel the need to get into it with you. You have the right to be angry at whatever you wish. I understand how much you want to defend Tesla and I think the CT is a great product in general. But every product has some issues. I'm just sharing my experience and I see others doing the same now that I have looked a bit deeper. I hope FSD improves quickly but the gap to actual full self-driving seems to be bigger than I and some others would have expected given the anticipation and the amount of years that have gone by now. I'm sure if you have used FSD for years, you probably have seen leaps of improvement but as a newcomer, it's surprising how much more there is to go. No shade is intended, just honesty.

I don’t think anybody here is angry.

There is a reason that “supervised” is being stressed. It is not “unsupervised.”

Compared to the C++ program version with over 300,000 lines of code, the latest FSD is head and shoulders better. And it is LEARNING.

I’m glad that you decided to eventually supervise your truck and get it off the light rail path. It probably should not have been permitted to even venture that far. Giving it that much leeway and freedom is just too much.

3-4 years ago I found a location in North Phoenix where FSD would literally drive your car into a median and then, based on speed and direction, what could be assumed to be into the path of opposing traffic. It was a repetitive glitch that I tested many times. But I never allowed the car to actually DO it.

The future is not quite here. But the newest neural-network for training FSD is clearly a correct pathway. Having had years of experience testing it, the neural-net version that I tested on my Model 3 convinced me to pull the trigger on the CT because Tesla has finally figured it out.

Wait for 10-10-24!

Sponsored

 

[dalton108]

I don’t think anybody here is angry.

There is a reason that “supervised” is being stressed. It is not “unsupervised.”

Compared to the C++ program version with over 300,000 lines of code, the latest FSD is head and shoulders better. And it is LEARNING.

I’m glad that you decided to eventually supervise your truck and get it off the light rail path. It probably should not have been permitted to even venture that far. Giving it that much leeway and freedom is just too much.

3-4 years ago I found a location in North Phoenix where FSD would literally drive your car into a median and then, based on speed and direction, what could be assumed to be into the path of opposing traffic. It was a repetitive glitch that I tested many times. But I never allowed the car to actually DO it.

The future is not quite here. But the newest neural-network for training FSD is clearly a correct pathway. Having had years of experience testing it, the neural-net version that I tested on my Model 3 convinced me to pull the trigger on the CT because Tesla has finally figured it out.

Wait for 10-10-24!

The one guy in SoCal (I believe he was an engineer) died several years ago doing the exact same thing! Except, unlike you on the day that he died he refused to take control of the car. basically, playing some kind of twisted game of chicken with the car. This was a very early auto pilot build if, I remember correctly.

His estate/family sued Tesla and they lost. It came out at trial that the car had repeatedly failed at this location prior to the fatal incident and the driver knew it! This was not the fault of the vehicle the decedent was the only person driving the car!

He was no different than the wife in “Memento” letting her husband inject her with insulin again and again until she died. It’s a tragic incident but it only highlights the unassailable fact that we are personally responsible for the operation of these vehicles!

 

[evnow]

As it prepared to make a left turn at a stoplight that was red, the left turn signal activated as we approached the intersection, but for some reason we stayed in the middle (go straight) lane even though there was a clearly defined left-turn lane.

FSD has always had this lane selection issue. You have to watch and disengage. That is where the supervision comes from.

If it does something that is clearly wrong, you have to intervene.

Average intervention rate is something like once every 20 miles in the city. So, you should expect to take control once or twice a drive - and if you are new to FSD, probably more since you would be unsure when to intervene. Better to intervene when not necessary than the other way round.

 

[Mark Shaw]

My first reply didn’t make it as someone thought it was political. It wasn’t. What I said just like different people think differently- there are many knuckleheads in life. Since some believe FSD is not good enough than I say - you didn’t pay for it - so if it isn’t good enough- don’t use it and perhaps pay to educate your children or grandchildren and MAYBE they will be able to improve a future version that you believe- is good enough for you b

Sponsored