A new article from the series:

Tuesday, January 8, 2013
*Universal Artificial Intelligence - Matches of Todor Arnaudov's teenage
works from 2001-2004 to later (2005-2006-2007) related more popular works
by Marcus Hutter and Shane Legg, Additional Discussions and other Todor's
works, and a new Challenge/Extension of the reasons for Herbert Simon's
"Bounded Rationality" (Satisficing)
*
http://artificial-mind.blogspot.com/2013/01/universal-artificial-intelligence.html

Contents
1. "Nobody knows what intelligence is"...
2. Definition of Machine Intelligence
3. Supposed Reasons for Behaviorism Avoidance by the Psychologists Community
4. Introduction of the Principles of Human Behavior Drive - More Pleasure
and Less Displeasure for Maximum Possible Time
5. Observation, Reward, Action and Bounded Reward
6. Meta-Physical reasons for reward and punishment
7. Local maximum, biggest sum of pleasure for a given time ahead (...) - a
multi-topic paper from 2004, which implies an extension of the Bounded
rationality
8. Extension of the Bounded rationality - *there are no non-trivial global
undeniable single optimum external rewards for intelligent agents as
complex as humans* **(see note in the article)*
9. Physical vs Cognitive rewards
10. The period ahead should be short enough for the target resolution,
range, domain etc.
11. Defining Universal Intelligence, Turing test faults, better tests...
12. Memories and background of the author in 2001
Appendices:
- Original texts in Bulgarian of part of the citations
- Link to another short discussion on the matches between Hutter and Legg's
setting and my works, published in an article from 2/2010


Notice p. 8. - that's a concise expression of my reflections regarding:

- Confusions about the "irrationality" of human decision making
-  My claims that an integral self does not exist and believing that it
exists cause confusions in philosophy, ethics, psychology, ... for millenia
-  My notion of "Calculus of Self" - an Integral of self over a period of
time, see: *"Nature or Nurture: Socialization, Social Pressure,
Reinforcement Learning, Reward Systems: Current Virtual Self - No Intrinsic
Integral Self, but an Integral of Infinitesimal Local Selfs  - Irrational
Intentional Actions Are Impossible - Akrasia is Confused - Hypothesis about
Socialization and Eye-Contact as an Oxytocin Source"*

http://artificial-mind.blogspot.com/2012/11/nature-or-nurture-socialization-social.html

When a different set of virtual control-causality units could be in charge
of governing the body, that allows for different reward functions to
compute different possible scenarios which in turn can produce sets of both
equal or opposite results. That is made even worse by the bounds of rewards
- you can't be "infinitely happy" and infinitely satisfied, -  and the
multi-dimensionality of the rewards, due to the non-integral self and
sensory multi-modality and multi-neurotransmitter reward and punishment
systems. Different reward functions and different scenarios may also imply
completely different spatial and temporal resolutions.

I claim that choosing any scenario among all those is rational at the
moment of making the decision, for the chosen scenario and the believes in
this very moment of the "virtual self" governing the body in that very
moment.

See for example the story from that 2004 work about the child, chocolate
and the dentist  (links from the article).

A) If the child chooses to predict the sum of the expected reward for the
next second, this is a maximum reward, local maximum and the best decision.

B) If the child predicts the sum of the expected reward for the next 6
months, by constructing a virtual universe regarding chocolate eating in
long term, she may recall that last time when she visited the dentist, the
doctor told her that eating chocolate blocks may cause caries. The
procedure at the dentist was painful, and the child now instead of getting
excited, gets frightened and may make exactly the opposite decision -
eating the chocolate will cause pain, so it's a local minimum, then she
will not do it. *

Notice that any of both actions is "rational" and consistent in its
resolution/range and interpretation, and selection of any of the
resolutions and interpretations is right at the moment of decision making
and the information which is given to the function summing expected reward.

Besides, these both are just about eating one type of food - there are many
other possible actions and functions.

.....
* The principles of basic drives of human behavior, at least according to
my definitions and observations, are more for less displeasure/pain, than
to equal pleasure=pain

*No pain  > pleasure=pain*

In fact "pleasure=pain" is not precise, because they are multidimensional
individually, and are in different dimensions together. There are pairs of
emotions and pairs of neurotransmitters, which are "antidots" to each
other, such as oxytocin and cortisol, but physical pain from hurting the
body uses different mechanisms, and there's not really a single dimension.

"Pain" is a sense indicating possible damage, which may be irreparable,
while the abstract "pleasure" (reward) is indicating (in my definition) the
coordinates where the lower type of control-causality units of mind should
aim, it indicates to those units how close they are to their local maximum,
are they near to the coordinates of "what they should" do. On the other
hand, pain (punishment) indicates where they shouldn't be.

No pain indicates that they are somewhere in between, say 0, but in that
state they still may reach to the goal-state later, while the pain-state
may be a sign of approaching irreversible destruction, after which they
would be able to reach to "nirvana" after an infinite amount of time
(clocks, cycles).

The above is about virtual micro-units, which macroscopically and after
integration produce mind and behavior, which at appropriate low resolution
and by integrating infinitesimally can also be seen as such
control-causality unit.


-- 
....* Todor "Tosh" Arnaudov ....*
*
.... Twenkid Research:*  http://research.twenkid.com

.... *Self-Improving General Intelligence Conference*:
http://artificial-mind.blogspot.com/2012/07/news-sigi-2012-1-first-sigi-agi.html

*.... Todor Arnaudov's Researches Blog**: *
http://artificial-mind.blogspot.com



-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com

Reply via email to