[gurps] Failing autopilots and RVO

[Onno Meyer]

> > One other thing about computer software that might come into play is
> that it
> > is predictable.
> 
> The problem with that is it would not be true, if the program were to
> use the Monti Carlo to solve the problems. Remember that this gets and
> answer by trying lots of random answers and picking the best bet.

A couple of points on that:

* Hellburner by C.J. Cherryh is a nice science fiction novel which
  talks about that kind of stuff. Can the AI avionics override the
  human pilot when he tries 'suicidal' moves, or does that make 
  the starfighter too predictable?

* Random solutions are much harder to test and verify, and vehicle 
  control programs have much higher quality standards than games 
  or even accounting programs. 'That was a crit failure, one in a 
  million' is no good answer when a mechanically sound airliner 
  plows into the ground. 

* Legal reasons are stopping some real-world automobile robotics, 
  too. There has to be a human driver in charge ...

* Monte Carlo analysis cannot cover all possible actions by the 
  robot, only those which are rewarded by the scoring algorithm.
  A human bomber pilot might decide to activate the navigation 
  lights and set the transponder to a civilian frequency. The 
  stochastic algorithm would conclude that lights won't jam a 
  missile and reject this course of action.