4 ms·

Quote from this site: "Current Autopilot features require active driver supervision and do not make the vehicle autonomous."

by pps 8y ago

cromulent 8y ago
Sure, and it had supervision. But Tesla's software made the choice, and the supervisor overrode it, but only because he fortuitously had time to do so. If a passenger aircraft has an autopilot problem that causes a crash, do you blame the manufacturer or the pilots? The pilots are always supervising the plane, and it's not autonomous, right?
- bencoder 8y ago
  Judging by a lot of the discourse immediately following the most recent 737max incident, people were definitely blaming the pilots. Thankfully that's cleared up a bit now.
  - notahacker 8y ago
    I think that's a very relevant analogy. By the standards of the average driver, the 737 Max pilots were extraordinarily well trained and highly skilled, and they had a lot of time to correct the software error relative to most conceivable problems with a semi-autonomous vehicle. But regulators have still (reasonably) decided that their inability to cope with software quirks should be treated as a software problem and not a problem with the pilots' failure to rectify the erratic behaviour.
  - dsfyu404ed 8y ago
    You and me must have read very different discourse. Anyone suggesting that Boeing not be immediately burnt at the stake was not and is not received positively at least not on HN.
- JauntyHatAngle 8y ago
  >If a passenger aircraft has an autopilot problem that causes a crash, do you blame the manufacturer or the pilots? The pilots are always supervising the plane, and it's not autonomous, right? I actually agree with you, but this is not a good point. In many cases (not all, but many) pilots are definitely partly to blame for an auto-pilot failure causing a crash, as they operate on the assumption that there are many fail-safes, and the pilot is definitely one of them. A Tesla having a failure and a driver reacting to it is very similar to the pilot acting as a fail safe to a malfunctioning auto-pilot which definitely happens. Where it falls down, is that drivers are not a trained, responsible part of the fail-safes (in my view), you can't rely on them paying attention and reacting in time, partly as they aren't paid professionals with simulator training, and partly because they don't have the 30,000ft of airspace buffer to recover from and clear airspace.
  - erikpukinskis 8y ago
    How is that different than, you know... actual driving? Anyway, isn't this just a purely empirical concern? How many accidents happen to people who use autopilot, how many to those who don't. If the second number is bigger, autopilot is a safety feature. If it's smaller, it's a safety liability. End of discussion. No?

Quote from this site: "Current Autopilot features require active driver supervision and do not make the vehicle autonomous."

by pps 8y ago

Quote from this site: "Current Autopilot features require active driver supervision and do not make the vehicle autonomous."

cromulent 8y ago

Sure, and it had supervision. But Tesla's software made the choice, and the supervisor overrode it, but only because he fortuitously had time to do so. If a passenger aircraft has an autopilot problem that causes a crash, do you blame the manufacturer or the pilots? The pilots are always supervising the plane, and it's not autonomous, right?

bencoder 8y ago
Judging by a lot of the discourse immediately following the most recent 737max incident, people were definitely blaming the pilots. Thankfully that's cleared up a bit now.
- notahacker 8y ago
  I think that's a very relevant analogy. By the standards of the average driver, the 737 Max pilots were extraordinarily well trained and highly skilled, and they had a lot of time to correct the software error relative to most conceivable problems with a semi-autonomous vehicle. But regulators have still (reasonably) decided that their inability to cope with software quirks should be treated as a software problem and not a problem with the pilots' failure to rectify the erratic behaviour.
- dsfyu404ed 8y ago
  You and me must have read very different discourse. Anyone suggesting that Boeing not be immediately burnt at the stake was not and is not received positively at least not on HN.
JauntyHatAngle 8y ago
>If a passenger aircraft has an autopilot problem that causes a crash, do you blame the manufacturer or the pilots? The pilots are always supervising the plane, and it's not autonomous, right? I actually agree with you, but this is not a good point. In many cases (not all, but many) pilots are definitely partly to blame for an auto-pilot failure causing a crash, as they operate on the assumption that there are many fail-safes, and the pilot is definitely one of them. A Tesla having a failure and a driver reacting to it is very similar to the pilot acting as a fail safe to a malfunctioning auto-pilot which definitely happens. Where it falls down, is that drivers are not a trained, responsible part of the fail-safes (in my view), you can't rely on them paying attention and reacting in time, partly as they aren't paid professionals with simulator training, and partly because they don't have the 30,000ft of airspace buffer to recover from and clear airspace.
- erikpukinskis 8y ago
  How is that different than, you know... actual driving? Anyway, isn't this just a purely empirical concern? How many accidents happen to people who use autopilot, how many to those who don't. If the second number is bigger, autopilot is a safety feature. If it's smaller, it's a safety liability. End of discussion. No?