[martindbp]

> since 2014, and nearly a decade later they aren't that much closer

That is just an absurd statement. AlexNet came out in 2012 and ResNet in 2015 but somehow Tesla had something "close" to the current FSD in 2014? They're heavily using Transformers now which came out in 2017. They have demonstrated leaps in performance since the first version of FSD Beta two years ago.

[rvz]

> That is just an absurd statement.

No it isn't. It is also not early days either and we have given FSD enough time to meet the claims of achieving Level 5 with those strong and 'confident' deadlines even set by Musk himself, which he knew he couldn't meet, but the gullible customers and fans rushed in and fell for it anyway.

Not only they have all been missed, this year, Tesla finally admitted that it was Level 2 and it still doesn't work in the most dangerous time to drive which is at night. FSD (Fools Self Driving) is no where near close in the nearly ten years it has existed.

This is an obvious over-promising of a broken product with the pied-piper selling it to their fans scam.

[Geee]

Admitted? The difference between L2 and L5 is liability, it is not a metric of how capable the system is.

[rvz]

> Admitted?

Yes. Admitted. [0]

> The difference between L2 and L5 is liability.

Everyone from the regulators to Musk himself have known that for years that it is more than just liability and none of the claims of achieving 'Level 5' FSD and the hype around the robotaxis have ever materialized with many customers falling for the scam lead by the pied piper (Musk) and once again increased prices on a broken product.

[0] https://www.news18.com/news/auto/teslas-full-self-driving-cl...

[Geee]

The driver has to be liable, until the L5 system has lower error-rate than a human. For regulatory and liability reasons, it should be registered as an L2 until it's perfect. No one has ever claimed that FSD is a finished product.

[rvz]

> The driver has to be liable, until the L5 system has lower error-rate than a human.

Which that was promised to be completed in 2020. Where are the Level 5 robotaxis that Elon was confident enough to release by that year?

> For regulatory and liability reasons, it should be registered as an L2 until it's perfect.

And FSD was advertised as Level 5 by Tesla and was due to be complete by 2020 deadline. Even before that, many customers fell for this and were promised that future updates would eventually make it Level 5.

See how the scam works? Not only it was meant to be 'Level 5' as advertised by 2020, Tesla knew that it won't get there all along and continued the deception as their fans bought into the false promise and customers realised that they are paying for a broken contraption that didn't function as advertised and will continue paying for it as the price keeps increasing.

> No one has ever claimed that FSD is a finished product.

Yes. Elon (and Tesla) were the only ones that claimed and promised that it will be finished with robo-taxis on the road, at Level 5 FSD by 2020. This is the pied-piper scam and they have sold a false promise to their loyal fanbase and they will pay for an unfinished product until its 'perfect Level 5 FSD'; whenever that is and will continue to raise prices.

I also don't think it is ok to use unfinished safety critical software on the roads and put many driver's on the roads lives at risk.

[Geee]

FSD is a level 5 system, but it just has too high error rate to give it full control. It's a beta software for a level 5 system.

Level 5 means that the system is designed for autonomous driving in every situation. It's not a metric whether the system actually works. There will be multiple level 5 systems on the road some day, and some of them will make mistakes more than others. Just like all technology and software, it's impossible to make these systems perfect.

Yes, Elon overpromised and they failed to deliver in time. This was compensated by cheaper prices, just like is usual in beta programs. Today, FSD is already insanely good and can drive 99% of time without human intervention.

It's still not good enough, and it's multiple years late from original forecasts. Still, I wouldn't call it a scam. It's a delayed project with a pre-order/beta program, but given it's huge complexity and importance, I'm willing to forgive them. When it's ready, no one cares that it was a few years late.

[washbrain]

Yeah? It was a toy then, and it's a toy now.

Yes, it's a much better toy, but it's still a toy.