>From the Greek On, "being" as in ontology, OnML would have been called
ObjectML if the name was available, and is meant to serve as a store
for object identities and histories, and could be the base for
PersonML. In fact human cognition is so anthropomorphic that objects
could have been modeled in PersonML as well. It has to be said that
the notion of an object in good old OOP is not really that far removed
from the requirements of AGI, if you consider that there will be
entities in the world whose behavior is/would most concisely be
captured by "methods", ie something more dynamic than XML. The
intention here is to capture and convert as much as possible of the
dynamic information into readily available static descriptions, even
if it results in hidden complexity in the form of natural language.
Philosophical efforts such as "you cannot cross the same river twice"
have not really had much impact on everyday life, but there does come
a time when we are forced to declare "he is not the same person
anymore" or "this is not the forest of my childhood". I don't think we
have any theoretical results suggesting one way to represent such
facts is better than the other, but it does seem reasonable to
represent the most salient versions of objects as Object1, Object2,
Object3 etc. Most people alive would not mind at all if assassination
attempts against Hitler were successful during WWII, but I am guessing
they would raise some objections when he was 25, 15 or 5, in other
words there is always a Hitler1 and a Htiler2. The idea is that for
understanding and reasoning you would like to process Hitler's
biography once and explicitly represent different epochs for future
reference, rather than scan the PersonML description (or any other
data store) time and time again.

Speaking of AGI and in light of the circular discussions we are having
most of the time, I would like to coin the term Artificial General
Human Intelligence, AGHI. I don't think this should be half as
controversial as Ben fathering AGI from AI. If you just consider the
biases and suprastructure involved in something like friendly AI, it
is further apart from AGI, compression, AIXI than AGI is from
alpha-beta pruning and mini-max. As an example, I would assert that
OpenCog's current architecture where notions such as "is" and "has"
are evolved will have a hard time converging to a human narrative
where some things, some of the time, "are" and "have" in very rigid,
symbolic ways. It is the same objection one would raise to
connectionism, only much worse.

Finally, the Japanese events remind us that high intelligence, high
density, low entropy, high interdependency are all accidents waiting
to happen. Smart as we are, we have to progress *intelligently*


-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/8660244-d750797a
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=8660244&id_secret=8660244-6e7fb59c
Powered by Listbox: http://www.listbox.com

Reply via email to