Would you take a ride in a driverless car?

[RingOfFire]

Sure! You only live once. I mean, well... okay but that once is pretty important. But yeah, I'd do it! What the hey!

NO WAY! I'd rather have pins stuck in my eyeballs then get in one of those death machines!

[MrOracle]

While there's still a way to go, I think MOST people have NO IDEA how close Tesla is to having a real, fully-working product.

Previously, their FSD (full self-driving) Beta software was hand-coded by programmers, and it was not bad, but it could also be very "robotic." All versions of FSD through v11 were this hand-coded software.

For FSD v12, that hand-coded software was completely thrown out, in favor of AI-based code. The AI is "trained" by analyzing videos of GOOD drivers out in the world correctly handling all of the weird situations that come up in the real world. And because there are millions of Tesla cars driving around in the world, Tesla has orders of magnitude more data than anyone else. In fact, Tesla's biggest problem right now is a lack of computing power, despite huge orders of NVidia AI processors that they've got online. Tesla has also designed their own AI processor and is having them produced, but they don't have nearly enough of them yet - a problem that is rapidly being solved. These super-computers will do nothing except analyze driving videos and "learn" the right things to do in various situations.

FSD is *still* a Beta product - it's still in testing and not yet "good enough" to be used completely unattended - but it's startlingly close and it's improving daily as problem areas are found and more videos of that problem are fed into the super-computers. It's really only the rare "edge cases" that are still sometimes a problem, and that's what is being worked on. The pace of this work improves as more computing power comes online, allowing the AI to process videos faster. We're definitely in the "home stretch", and I suspect that 2025 will be the year that FSD leaves "beta" and becomes a real product.

https://www.youtube.com/embed/1hSjfmBgI8g

As to the other self-driving efforts - they're a joke. They barely work, and they only work in very specific, relatively small areas that are pre-mapped with high-precision LIDAR. It's kind of like the little cars at Disneyland that run on a track - they can only work on their little fixed area, and even then, they require $100,000 of sensors bolted on to whatever car they're using, which isn't even remotely scalable. Meanwhile, a brand new $38,000 Tesla Model 3 has all the sensors and processors it needs already on-board. There's really no comparison.

[RingOfFire]

Call me back when there are millions of them on the road and their accident rate is equal to or less than cars with drivers in them. I'll believe it when I see it. As of now I can't even get through a day without some kind of computer issue.

[RingOfFire]

PS giving you a thumbs up for your optimism.

[MrOracle]

If they were to turn on FSD *today*, in it's unfinished state, it would ALREADY radically reduce accidents and deaths. In fact, if every car had it - even as it is today - accidents and deaths would be nearly eliminated. I'm not sure most people realize just how AWFUL the average HUMAN driver actually is. Improving on humans really isn't a high bar to cross.

Tesla is looking to improve human performance by a couple orders of magnitude. This isn't going to mean that accidents will be completely eliminated, especially at first, because there will still be way too many humans on the road, with some percentage of them doing really stupid things. And the media will blame Tesla and FSD at every opportunity, because "Elon bad." But over time, the truth won't be able to be denied, and more and more people will demand it. It's very likely that Tesla will eventually license FSD (with a number of specific requirements) to other OEMs, because none of the other attempts are even remotely close.

[MrOracle]

I forgot to mention that an investigative reporter actually ran the numbers a year or so ago, back on FSD v11, and found that human drivers got into accidents at a rate of 16 times more often than FSD v11 did. Further, a number of FSD v11 accidents involved drivers intentionally circumventing the safety requirements, causing NHTSA to require that Telsa increases driver monitoring. There has yet to be a single death when FSD has been in use. There have been several stories in the media speculating that FSD caused deaths, but they were all proven false (FSD was not, in fact, in use).

[MrOracle]

You might find this YouTube channel interesting:

https://www.youtube.com/watch?v=dntws31nAc4

[exitseven]

Yes, I can't wait until we ha e driverless cars. i will just sit and watch a movie or something.

[RingOfFire]

What movie? One about a Demolition Derby? 🤣

[exitseven]

Maybe Vanishing Point?

[RingOfFire]

🤣
...

[exitseven]

Thanks for the MHO😎

[Phoenix98]

Of course not I don't trust the technology at all and I don't trust the people in charge of it either.

[msc545]

As long as I can grab the wheel and push the break when it all goes wrong.

[OtterMan68]

Depends. Where are we driving? How fast? What's the traffic like?

[EmmaMary]

No thank you very much.

[strateguy632]

great photos!

[Luckycharms12]

Nope

[Anonymous]

Maybe in 10 years.