From: Anders Sandberg (asa@nada.kth.se)
Date: Sun Mar 18 2001 - 01:54:54 MST
Nick Bostrom <nick@nickbostrom.com> writes:
> Robert Bradbury wrote:
>
> >However the other part of this "if life was all happiness, people would become
> >bored" raises interesting issues.
>
> That sounds like nonsense to me.
But it is a very common meme. I wonder how true it is; if pleasure and
pain were just a one dimensional scale and exact antonyms it would
likely be wrong, since any difference in valence would be just as
useful as the absolute levels. But there are reasons to think rewards
and punishments are asymmetric, and that might mean there really is a
need for both. Of course, this doesn't mean pain has any intrinsic
value, but a signal like that (with the necessary level of
specificity, aversiveness etc) would be useful.
Tragedies might be useful training for recognising complex (but
possibly common) dilemmas where the outcome is a lose-lose situation.
> In any case, we can say with absolute confidence is that boredom is not
> going to be a problem. Boredom is an emotion and already we have means
> (drugs etc.) that can reliably dispel boredom for extended periods (albeit
> sometimes with side-effects). It is unthinkable that a jupiterbrain SI
> wouldn't be able to figure out a way of avoiding the boredom emotion if it
> chose to.
Did I mention that one of my students implemented a kind of boredom in
his neural network agent to make it get out of loops? After repeating
itself too much the "boredom/frustration" (implemented in a rather
clever low-level way) would build up and make another behavior more
likely. I think this is the reason why a certain propensity for
boredom is useful. The problem is tuning it: if the buildup is too
slow, you become obsessive, if it is too fast you jump from project to
project.
-- ----------------------------------------------------------------------- Anders Sandberg Towards Ascension! asa@nada.kth.se http://www.nada.kth.se/~asa/ GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:06:29 MST