Dennis
1) I agree that direct reward has to be in-built
(into brain / AI system).
Okay.
2) I don't see why direct reward cannot be used for rewarding mental
achievements.
They could be, dependant upon the type of system you are interested
in. Not easily the one that I am interested in.
I think that this "direct rewarding mechanism" is
preprogrammed in genes and cannot be used directly by mind.
This mechanism probably can be cheated to the certain extend by the
mind. For example mind can claim that there is mental achievement when
actually there is none.
That possibility of cheating with rewards is definitely a problem.
I think this problem is solved (in human brain) by using only small
dozes of "mental rewards".
For example, you can get small positive mental rewards by cheating your
mind to like finding solutions to "1+1=2" problem.
However, if you do it too often you'll eventually get hungry and would
get huge negative reward. This negative reward would not just stop you
doing "1+1=2" operation over and over, it would also re-setup your
judgement mechanism, so you will not consider "1+1=2" problem as an
achievement anymore.
I don't like having a judgement mechanism of the complexity you
describe. How does it determine the pointful repetitive tasks
(job/scientific research) from the pointless (1+1=2)? Both of which
may force you to avoid food. Is it also in-built?
Also, we all familiar with what "boring" is.
When you solve a problem once - it's boring to solve it again.
I guess that that is another genetically programmed mechanism with
prevents cheating with mental rewards.
3) Indirect rewarding mechanisms definitely work too, but they are not
sufficient for bootstrapping strong-AI capable system.
Consider a baby. She doesn't know why it's good to play (alone or with
others). Indirect reward from "childhood playing" will come years later
from professional success.
You neglected the respect from peers as direct reward. I should really
have been more specific and said that we enjoy when we get positive
body language from people around us. This is how we teach children, a
child walks/talks they gets positive body language from adult, thus
does more of it. Playing with other kids gets us some positive body
language from the other kids and also enables us to learn how to do
more things and thus get more positive body language from parents or
find out how to get sweets.
Baby cannot understand human language yet, so she cannot envision this
success.
AI system would face the same problem.
I think you might mean something different by indirect reward that I
do. I am not at all talking about envisioning success or anything to
do with language.
My conclusion: indirect reward mechanisms (as you described them) would not be
able to bootstrap strong-AI capable system.
Who knows, until I have done some proper experiments I can't say here nor there.
Back to real baby: typically nobody explains to baby that it's good to play.
But somehow babies/children like to play.
My conclusion: there are direct reward mechanisms in humans even for
things which are not directly beneficial to the system (like mental
achievements, speech, physical activity).
If you consider spreading your genes to be the "goal" of a human, then
having a one night stand with a condom on is not directly beneficial
to the system. So I would agree with you. I can accept any direct
reward mechanism as long as you can plausibly argue that the mechanism
for the state for giving the reward is genetically specified. I have
trouble with a finished crossword genetically specified reward.
Possibly you could posit some busy brain finishing reward, but as far
as I can tell this is still too open to exploitation.
Will Pearson
-------
To unsubscribe, change your address, or temporarily deactivate your subscription,
please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
