SF Fire Dept.: Person Dies After Cruise Robotaxi Blocked Ambulance

Post by ebalit » Sun, Sep 03, 2023, 7:46 PM UTC

flandish wrote: post #102

joshribakoff wrote: post #87

Earlier quoted context omitted.

The only way this will happen (anytime soon) is if they put the safety drivers back in but they will never do that willingly because they have worked for a decade and invested billions to get to this point and it would be difficult to capitulate. There is simply no way to remotely assist reliably in the same way as being physically present (no network has 0 latency and 100% reliability). The AI likely cannot drive “p…

For sure. Us FD (and some police) have what we call “knox box keys” - knox is a brand. They’re locked in the apparatus, and with a pin code or radio signal I can get them unlocked and then from there unlock a little box on a building. That contains the buildings master keys. Anyway - if apparatus contained corporate keys (fobs, fido, physical, whatever) that allow a FD member to drive the car - I can 100% see that be…

Post by tremon » Sun, Sep 03, 2023, 8:00 PM UTC

halfcat wrote: post #111

falcolas wrote: post #89

Earlier quoted context omitted.

If you shoot a gun out your window and kill someone who jumped off the building, you are still on the hook for manslaughter. That is to say, Cruise is still on the hook, even if they didn’t hit the pedestrian in the first place.

What if a brand new, human-driven taxi breaks down, blocking the ambulance? Is the taxi company on the hook?

Post by rich_sasha » Sun, Sep 03, 2023, 8:06 PM UTC

perihelions wrote: post #62

- "blocked a San Francisco ambulance from getting a pedestrian hit by a vehicle" I'm taking no "position" on either "side" of the debate, but I feel it's noteworthy that a *human-driven* car killed a pedestrian (seriously?) here, and that fact is utterly banal and completely unremarkable to the authors—and to us. AI cars are still highly flawed and kill people. Human-driven cars are highly flawed, probably more flawe…

Post by CamperBob2 » Sun, Sep 03, 2023, 8:22 PM UTC

mcpackieh wrote: post #103

CamperBob2 wrote: post #33

Earlier quoted context omitted.

Human drivers kill 30,000+ per year in the US alone. Development of an automated driving option is a safety-critical context, and that requires real-world deployment experience. The precautionary principle is not what is called for here.

Your argument has the form of: "We have to do something. This is something, therefore we have to do this." You've failed to establish automated driving as a safer alternative, let alone the best practical solution. The fact that these cars can't even move out of the way of emergency vehicles proves they aren't ready for testing on the public now. > requires real-world deployment experience. Not at this stage it doesn…

Post by CamperBob2 » Sun, Sep 03, 2023, 8:27 PM UTC

joshribakoff wrote: post #67

CamperBob2 wrote: post #33

Earlier quoted context omitted.

Human drivers kill 30,000+ per year in the US alone. Development of an automated driving option is a safety-critical context, and that requires real-world deployment experience. The precautionary principle is not what is called for here.

I believe you’re citing a stat corresponding to the number of traffic fatalities. That doesn’t mean in all of those cases one person killed another as you implied. For example if someone dies after drunk driving and crashes themselves into a tree that doesn’t seem right to say “the driver killed someone”