171–180 of 850 posts
GM's recent TV advertisement shows the driver engaging their version of autopilot and crossing his hands...I was like woah! That would never make it through legal if they weren't very very confident about their tech. Contrast that image with Tesla's demand that drivers keep their hands on the wheel and it seems that GM is miles ahead of everyone on autopilot.
The fact that the driver received disengagement warnings prior in the drive is not very relevant to the accident per se. Might as well say the driver has received multiple disengagement warnings during his ownership of the Tesla. It is not relevant to the actual circumstances of the accident. Note that they do not say "he received disengagement warnings right before the accident and chose not to respond". Tesla instead says that his hands were not detected on the wheel 6 seconds prior to the accident. Okay, so thats the case, but obviously the Autopilot system allows that to happen, as disengagement warnings were not active at that time, therefore THE SYSTEM WAS OPERATING WITHIN ITS ESTABLISHED BOUNDS.
This leaves it down to, the autopilot system drove into the barrier while the driver wasn't paying attention, but was operating the vehicle within the defined bounds of the semi-autonomous system, but violated the parameter of "pay attention to the road and be ready to take control". He did not receive any system warning before the accident, he just simply chose not to correct the system behavior before the accident happened. Everything else is just trying to kick up dust and obfuscate the facts of the situation it seems.
The issue with Tesla's semi-autonomous system and to a certain extent, other systems, is that there is a poor boundary between system responsibilities and user responsibilities. This causes issues around the boundary where sometimes the autonomous system will accomplish something, but sometimes it can't and the user must. Because of this fuzzy boundary, I think its almost more important to invest into disengagement logic, equipment, and programming than pure self-driving technology. A good example of investment into this arena is actually GMs Super Cruise, which seems to have MUCH better user engagement detection and much sharper boundaries than systems like Tesla's
GM's recent TV advertisement shows the driver engaging their version of autopilot and crossing his hands...I was like woah! That would never make it through legal if they weren't very very confident about their tech. Contrast that image with Tesla's demand that drivers keep their hands on the wheel and it seems that GM is miles ahead of everyone on autopilot.
Given the bad press that Tesla gets after any Autopilot driven accident, GM seems to have made a safe choice here.
Blaming the autopilot for the death of Walter Huang is like blaming the bullet for a firearm-related death. The man at the wheel or the man at the trigger is ultimately responsible.
Because I don't see much difference from how Elon Musk makes misleading and clearly false statements.
Earlier quoted context omitted.
> STILL took his hands off the wheel That's not clear. The car didn't detect his hands on the wheel, but that doesn't mean his hands weren't on it. Tesla's implementation of hand detection (like many of their other systems) is rather unreliable.
So he had his hands on the wheel and took no action for 6 seconds while driving straight to the barrier. Doesn't make it look any better for the driver TBH.
Earlier quoted context omitted.
If Tesla was shipping a Level 3 system, your analysis would be great. But they aren't. They're shipping a Level 2 system where you're reminded each time that you enable it that you've got to remain alert. And which dings at you if you don't keep your hands on the steering wheel.
This is irrelevant. Before his death, Tesla told the driver that there was no issue with the autopilot feature. This likely contributed to his continued use of the feature. That feature, because of the same issue (which Tesla denied existed) killed him. And now Tesla is saying that there is an issue, and that it's his fault for using the system. Whether or not it's a full self-driving mode, or if it's an assisted aut…
Earlier quoted context omitted.
At the heart of this issue is the fact Teslas are the first car that can have its functionality changed drastically from day to day (this "drive into a median" bug is believed to be a recent issue introduced within the last few months of autopilot updates). I think that ability to update is one of the most exciting things about Teslas compared to other cars, but it presents a huge organizational problem. How does Tes…
As someone elsewhere in this thread put it, this is a case where silicon valley's posterchild slogan "move fast and break things" has turned into a reality of "move fast and end lives". Fast updates can be a great thing, but companies (in this case, Tesla) need to do a better job of figuring out when to apply this mentality of "move fast" and when to instead apply a mentality of "move slower because even the slightes…
They do and I'm largely glad they do but I think you can go too far the other way, delaying advances that can save lives can result in more lives been lost because of the delay.
There has to be a middle point, you want new drugs (for example) reaching the market but with the risk minimised to a degree that is acceptable, you simply can't reduce the risk to 0.
All that said I'd be fascinated to read the assurance approaches that Tesla uses before pushing out updates to what is essentially a computer on wheels attached to batteries.
Blaming the autopilot for the death of Walter Huang is like blaming the bullet for a firearm-related death. The man at the wheel or the man at the trigger is ultimately responsible.
Earlier quoted context omitted.
This is irrelevant. Before his death, Tesla told the driver that there was no issue with the autopilot feature. This likely contributed to his continued use of the feature. That feature, because of the same issue (which Tesla denied existed) killed him. And now Tesla is saying that there is an issue, and that it's his fault for using the system. Whether or not it's a full self-driving mode, or if it's an assisted aut…
You can't just blanket say everything is irrelevant because it doesn't fit your narrative. If he reported the system wasn't working, why would he continue using it especially in a known problematic area?
As for the relevance: sure I can. My post about Tesla's contradiction has no relevance on what kind of autopilot it was. Tesla denied there was an issue, contradicted themselves by then saying there is an issue, and now is casting blame on the user for that issue. That's not acceptable. It would be unacceptable if we were talking about windshield wipers, or tires, or door locks, or the fucking clock on the dashboard. It is all about Tesla's denial of the issue and contradiction of themselves, and the type of autopilot is completely irrelevant.
Earlier quoted context omitted.
At the heart of this issue is the fact Teslas are the first car that can have its functionality changed drastically from day to day (this "drive into a median" bug is believed to be a recent issue introduced within the last few months of autopilot updates). I think that ability to update is one of the most exciting things about Teslas compared to other cars, but it presents a huge organizational problem. How does Tes…
As someone elsewhere in this thread put it, this is a case where silicon valley's posterchild slogan "move fast and break things" has turned into a reality of "move fast and end lives". Fast updates can be a great thing, but companies (in this case, Tesla) need to do a better job of figuring out when to apply this mentality of "move fast" and when to instead apply a mentality of "move slower because even the slightes…