Comment by cherioo | Hacker News Reader

Comment by cherioo | Hacker News Reader

People always lament about how Tesla tries to mislead customer's about its capability, but is there actually data that shows this is truly the case among Tesla owners? How many times does "attention needed" beep is needed before customer can be considered reasonably informed that it is not fully self-driving?

Even among non-Tesla cars, whose manufacturers don't try to "mislead", 71 percent of people believe automatic emergency braking can avoid all crashes [0]. Is this percentage higher or lower for Tesla, and is that difference warranted?

[0] https://newsroom.aaa.com/2016/08/hit-brakes-not-self-braking...

[1] Quote from above "When traveling at 45 mph and approaching a static vehicle, a scenario designed to push systems beyond the stated limitations, the systems designed to prevent crashes reduced speeds by 74 percent overall and avoided crashes in 40 percent of scenarios. In contrast, systems designed to lessen crash severity were only able to reduce vehicle speed by 9 percent overall."

These cars are driving on public roads, so the question is not whether the user is informed but whether they are compliant. It doesn't take many ignored beeps to learn that a particular customer is not using the feature as intended, and deactivate it for the safety of others. Offer to turn it back on after watching a trainig video at the dealer and signing some papers once, escalate to a mandatory counselling session after that, third strike and the feature is permanently off. Put it in the TOS so that asshole drivers can't sue and win in court. The customer should not be king when the safety of others is on the line.

So speeding is the manufacturer’s fault too? Informed by the car, non-compliant driver...

Oh, and Tesla does disable Autopilot if you ignore the hands-on-wheel warnings.

Tesla advertises their cars as having full self driving... hardware. Even if this is true[1], it's pretty misleading to advertise their cars as having "full self driving capability" because they have the _hardware_ for it. The problem is clearly that the software isn't ready yet.

[1] I think it's pretty debatable, and as far as I know it's still something of an open question to determine the minimum amount of hardware required for a fully autonomous car

> How many times does "attention needed" beep is needed before customer can be considered reasonably informed that it is not fully self-driving?

I mean there is a warning telling you that it's not self driving. People are dumb, that's it.

I don’t think it’s helpful or even correct to say “people are dumb,” as if this is purely a user problem.

The system exists in a bad gray zone — one that works well enough most of the time you stop paying attention. But it’s not full proof, so you get the worst of cognitive bias here.

Personally I think this faux self driving (literally falsely named auto pilot) should be outlawed. We cannot depend on people to take over when the car generally is handling the situation most of the time. Either manual or fully automatic — this in between will continue to cause too many bad accidents.

Agree, if you are not actively driving, your attention, and therefore reaction time, will inevitably deteriorate. In addition to that your reaction time is further reduced by the fact that even if you anticipated a problem looking at the road, you need to wait long enough after when you expected the car to react before realising that the car didn’t react and you need to take over, which further reduces your reaction time. Which is why I have some sympathy for the drivers of experimental self driving cars. Doesn’t excuse them playing with their smartphone at the wheel but they are tasked with a really hard exercise while being legally liable if things go wrong.

How do we make the eventual evolution from faux self-driving to full self-driving if not through incremental improvements?

I'm far from an expert on the subject, but I assume real data must be acquired for that to happen.

You raise the bar of what you need to deliver which will change your approach, and don’t ship products that aren't ready.

It seems to me that delaying self-driving cars for any appreciable amount of time will lead to more deaths/injuries, not fewer.

You're assuming that the quickest route to safest auto-driving lies via incremental improvements to existing systems and that's not necessarily true; see: AlphaGoZero

Fair point. I think a game with specific rules is probably better suited to the self-learning process that AlphaGo uses, whereas real-world driving rules are much more complex and nuanced. But I think it's still something worth considering.

Reply on news.ycombinator.com