[MUD-Dev] TECH: Complex NPCs

Dave Rickey daver at mythicentertainment.com
Mon Apr 1 15:35:39 CEST 2002


From: Acius <helpsfamily at attbi.com>

> I've tried some stuff myself, but it's not too far yet. Let me
> start by saying this is a HARD problem. Saying "the technology is
> out there" is a bit naive. There is no well-understood, robust
> method for creating convincing artifical life (repeat three
> times). The problem is far more in the domain of psychology than
> computer science, and as such cannot be solved with ones and zeros
> alone. There's a lot more to rant about on this topic, but I won't

Depends on your definition of "convincing".  It's fairly easy to
create emergent A-Life simulations that behave generally as you'd
expect, with herbivores, carnivores, flocking and pack-hunting
behaviour, etc.  And it is possible (but far from easy) to create
systems based on neural nets that display personality and
distinctive habits that they have learned (Creatures and B&W being
by far the best demonstrations of this in a "game" context).

And the latest round of FPS's can do a very good job of simulating a
enemy, in fact playing Rabbit in T2 against the bots is more fun
than playing it against humans (since bots don't get bored and start
dropping out or initiating votes to change the game type).

But passing the Turing test and creating a game character with
"real" presence?  Not even close.

--Dave

_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev



More information about the mud-dev-archive mailing list