Hacker News Reader

Comment by Xylakant | Hacker News Reader

This will not help. Humans are fundamentally incapable of babysitting repeptitive boring tasks that require interaction only in failure cases. Expecting humans to do so is expecting superhuman capabilities and no amount of threatening the drivers with legal consequencces will change the outcome. All you'll achieve is punishing the poor sobs that trusted teslas marketing stunt. If you want this to change, make Tesla liable and they'll quickly fix this, one way or another.

View on HN

Making Tesla criminally liable (in a tangible, punitive sense, not just a $5000 fine per infraction) for every instance of an individual being directly harmed by a Tesla's self-driving mode would be a huge win for society as a whole, but Tesla and their supporters and apologists seem to have this idea that "hey, it's software, it's buggy but we'll fix it and iterate and then it'll be great" without realizing that we're not just crashing and losing the last ten minutes of someone's essay, we're crashing and killing human beings.

It would be an uphill battle for sure, and that's incredibly depressing.

The autopilot is adaptive cruise control + lane keeping. It’s abundantly clear and on top of this plenty of other cars have both features. The only difference is that the wheel is kept more steady.

Why would you make the manufacturer liable for this?

> As the recall mentioned above highlights, it was too easy for drivers to leave Tesla's Autopilot on its own and completely relinquish control.

(From the top post). If other manufacturers show the same flaw, make the liable, too.

For marketing it as „autopilot“ and not „adaptive cruise control“ like every other manufacturer.

And yet airliners fly across the world constantly without crashing into things. That's literally the job of an airline pilot . . . babysitting repetitive boring tasks that require significant interaction only in failure cases. I was military aircrew, so I know this. Sure actually operating the aircraft is a hell of a lot more complex than a car, but there are also fewer safeguards than they're talking about putting in cars to "keep people's attention." This doesn't require superhuman skill.

Perhaps the pilot licensing/training process filters out those who can't babysit? I would assume though that this versus the car licensing/training process (low bar of required training) is where the difference lies?

Apart from the training and the fact that pilots work in pairs, there’s a host of differences that make a planes autopilot task and behavior fundamentally different from what Tesla misleadingly labels autopilot:

Planes operate in a space with good visibility, any obstacle will be visible with ample lead time.

Planes operate in controlled space, with dedicated and verified spacing between traffic participants, allocated by an external traffic control. Collisions are not rare because pilots jump in within split seconds to avoid them, but because traffic control routes planes away from each other if they converge.

The planes autopilot is in fact less capable than what the teslas autopilot pretends to be. Its limitations are fairly well understood by pilots.

A planes autopilot doesn’t dump the pilot into a “you’ll crash in a second” situation. Handover from autopilot is a different process than what Tesla expects. The planes systems are designed with human psychology and limitations in mind, because of the comparatively high regulatory bar.

You gotta be pretty unlucky to "crash into things" while in the air with autopilot at cruise altitude.

Autopilot on cars is a MUCH harder problem than on an airplane. Plus you're taking about trained professionals and liable companies vs average Joe and negligent developer.

Different contexts and not really comparable.

1) The whole airspace environment is heavily regulated (ATC, rules, etc), cooperative (think transponders or TCAS for example) and usually quite sparse (several miles between planes).

2) Pilots are regularly trained and evaluated for these kind of take over in case of emergency.

3) You have two pilots monitoring each other.

4) The development cycles for these automations are far longer, and the QA a lot stricter.

5) At least in the most critical phases (take-off, landing), the automation is not exactly end-to-end. It's more the case of grouping batch of steps together into bigger actions to reduce the load on the pilots. The pilots are still quite busy during these periods and very much in the loop.

Punishing “poor slobs” is the cornerstone of deterrence for most crimes, why doesn’t it apply here as well?

It depends on what you want to achieve.

If all you want is revenge, then by all means, punish them as hard as you wish. Make them listen to tesla autopilot advertising until their ears bleed. The harm is done by that time.

If you want to avoid harm, then you're well served to take the failure mode into account - it's not a conscious decision to take attention away from the autopilot. It's a fundamental psychological mechanism and we're all prone to fall victim. The only way to avoid the trap is to not use autopilot - or change how it works and adress the need for changes to the party that (mis)designed the feature and has power to change it.

Humans are fundamentally incapable of babysitting repeptitive boring tasks that require interaction only in failure cases.

Read this repeatedly until you've answered your question.

Reply on news.ycombinator.com