27 ms·
> What's you're point? I'm pretty sure the point is that designing software to run in specific conditions is different than designing software capable of runni
by kbenson 8y ago
> What's you're point?
I'm pretty sure the point is that designing software to run in specific conditions is different than designing software capable of running in all conditions. This isn't just about the driver, it's about other people on the road and pedestrians. Just because the autopilot is sufficient to drive the car under the constraints the car ships with doesn't mean it's sufficient if the motors are putting out significantly more or less power, and its ability to detect this if it exists is somehow compromised.
Tightening your lug nuts is fairly straightforward and easy. Verifying physical functionality of an AI driving system is not. The only safe way to react if it can detect the system is not as it should be is to disable itself and require manual driving. It should do this in the case of standard components operating outside of acceptable specifications as well.
- jayar95 8y agoHonestly, autopilot software should have checks and balances already in place to prevent use if the system is compromised. Regardless if it was compromised by human hands or general use / wear & tear
- linuxftw 8y agoMy sentiments exactly.
- rootusrootus 8y agoTesla is hardly the only company that makes a car with lane following and adaptive cruise control. How do the other manufacturers deal with this issue?