The marketing is certainly a big part of it, but I think it's a mistake to think there aren't significant issues here that would remain even with better marketi

The marketing is certainly a big part of it, but I think it's a mistake to think there aren't significant issues here that would remain even with better marketing. One big problem is the sheer unpredictability of what's going to happen the next moment. If you turn on cruise control on the wrong street, its behavior is still going to be entirely predictable. You're simply not going to get suddenly erratic behavior just because you pressed the cruise-control button. Autopilot on the other hand seems to actively make the situation far more unpredictable than it would otherwise be, increasing the dangers people find themselves in. That's just horrible by any standards I can consider sane, no matter how it's marketed. Another (somewhat related, somewhat distinct) problem is that a system that works (say) 99% of the time and fails 1% of the time can easily be more dangerous than one that works 20% of the time and fails 80% of the time. (Obviously I made up the numbers, but adjust accordingly.) If you have to constantly correct something, then you're used to it; it's second nature to you. But if you might get used to it working too well... then as a human you won't be necessarily ready for the rare cases, especially when they're more difficult to predict. So an autopilot that works 'better' than something like cruise control can easily end up being more dangerous because of this, and it would be a mistake to hold them to identical standards.