[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Content
> > Wedding these kinds of content customizations to simulated environments
> > offers other possibilities. Perhaps a player might create and play a
> > character awhile, and then decide to leave that character in the world.
> > By using a rules based neural net learning system in addition to other
> > tables and systems such user-created denizens would be very dynamic,
> > shaping the experience of other players. And this wholly on the basis of
> > simulation.
>
> Agreed, but this makes two assumptions. First, that the
> designers are willing to trust the player to create part of the world,
> and thereby to shape the experiences of other players. Second, it
> assumes that the player has sufficient agency to create an interesting
> character. If the character is a generic potato with a bunch of
> bric-a-brac stuck on the outside, then an autonomous version of the
> character is essentially a monster with a name. I can think of few
> things more depressing than running into an NPC-lobotomized version of a
> former character, and finding he had been damned to repeat "Welcome to
> Arni Village!" for all eternity.
Indeed, this is why I mentioned neural net learning. This particular
approach to AI is very good at teaching a _dumb_ program to behave like
a player. Although a neural net driven AI would not in itself be driven
by personality variables or desires, as with characters in the
Erasmatron or The Sims, it would go through the motions of behaving as
the player would. Given any situation the player had encountered it
would do what the player had done, or choose from the most advantageous
optons among the player's actions. Given an advanced text parsing engine
a neural net could even learn to speak as the player had (although
"Welcome to Arni Village" might be preferable to "I lvl 12! I kill j00
sux0r!"). Neural net training might fall back upon 'bot' AI in new
situations or even assume standardized dramatic drives as a supplement
to the NN learned behaviour.
No conclusions,
--Bob