4 ms·

Self driving is one of the hardest, if not the hardest software problems ever to be solved. I think people underestimate just how challenging the problem is. Mu

Self driving is one of the hardest, if not the hardest software problems ever to be solved. I think people underestimate just how challenging the problem is. Must says a lot of stupid stuff but in reality he has delivered on the vast majority of it, though as he says it's not always in the time frame he expects.

skydhash 4y ago
Wouldn't something like Westworld road system easier? Basically, almost all the cars are self-driven, and it's more like a public transit than anything. Essentially, a closed system. (Easier in the technical sense, not the implementation (politics, ...)).
- TaylorAlexander 4y ago
  Even if that were easier, there’s basically no way to get there without full self driving capable of working around human drivers. The reason why cars are convenient is due to the existing road network, and any deployment of self driving cars is going to have to function on existing road. If you tried to segregate the system, you would have to close some of the existing roads but this would mess with regular people and result in a very small network of pure self driving roads.
itsoktocry 4y ago
>I think people underestimate just how challenging the problem is Is this the shifting of the narrative? "People" aren't underestimating the problem. It's widely known to be an extremely difficult problem. Musk said he'd have it done by now, and actually sold the technology. People have been skeptical of it for the better part of a decade.
- Robotbeat 4y ago
  I remain skeptical Tesla is anywhere near let-the-driver-sleep self driving (except perhaps in tunnels), but I’m very glad they’re trying to do it, and do it at scale for consumer cars, not just company-owned hyper-sensored vehicles that cost as much as a house. Literally every company working on the problem a decade ago was over-optimistic, but many pursued a strategy that would keep the technology only for company-owned vehicles, which I find less satisfactory. The whole point of personal cars is independence. We already have buses and transit vans if we want mass transit. If you want self-driving to actually save all the lives it supposedly could, it has to be scalable to everyone. Musk is terminally optimistic on AI generally, including paranoia about AGI. I don’t expect let-the-driver-sleep self driving for a good 10 years, but it’ll be never if no one tries. The technology does not advance by itself.
  - DoesntMatter22 4y ago
    People are already sleeping while the Tesla drives, albeit illegally. Highway driving it seems to do really well, around town though sleeping seems way off.
- DoesntMatter22 4y ago
  People underestimate it all the time, just because you don't doesn't mean other don't. If you read any forum they talk about how it should have been solved by now.

skydhash 4y ago

Wouldn't something like Westworld road system easier? Basically, almost all the cars are self-driven, and it's more like a public transit than anything. Essentially, a closed system. (Easier in the technical sense, not the implementation (politics, ...)).

TaylorAlexander 4y ago
Even if that were easier, there’s basically no way to get there without full self driving capable of working around human drivers. The reason why cars are convenient is due to the existing road network, and any deployment of self driving cars is going to have to function on existing road. If you tried to segregate the system, you would have to close some of the existing roads but this would mess with regular people and result in a very small network of pure self driving roads.

itsoktocry 4y ago

>I think people underestimate just how challenging the problem is Is this the shifting of the narrative? "People" aren't underestimating the problem. It's widely known to be an extremely difficult problem. Musk said he'd have it done by now, and actually sold the technology. People have been skeptical of it for the better part of a decade.

Robotbeat 4y ago
I remain skeptical Tesla is anywhere near let-the-driver-sleep self driving (except perhaps in tunnels), but I’m very glad they’re trying to do it, and do it at scale for consumer cars, not just company-owned hyper-sensored vehicles that cost as much as a house. Literally every company working on the problem a decade ago was over-optimistic, but many pursued a strategy that would keep the technology only for company-owned vehicles, which I find less satisfactory. The whole point of personal cars is independence. We already have buses and transit vans if we want mass transit. If you want self-driving to actually save all the lives it supposedly could, it has to be scalable to everyone. Musk is terminally optimistic on AI generally, including paranoia about AGI. I don’t expect let-the-driver-sleep self driving for a good 10 years, but it’ll be never if no one tries. The technology does not advance by itself.
- DoesntMatter22 4y ago
  People are already sleeping while the Tesla drives, albeit illegally. Highway driving it seems to do really well, around town though sleeping seems way off.
DoesntMatter22 4y ago
People underestimate it all the time, just because you don't doesn't mean other don't. If you read any forum they talk about how it should have been solved by now.