This is nonsense, Sergio. You're making it up - i.e. "there is only one
behavior" - to suit your argument, not from any observation of the real world.
Animals don't do just one thing or have just one response. (They wouldn't
dither, and oscillate as they do, if that were true). And what is at stake is:
(after you've defined their problem, [which you don't do], what are their
options - what options do and can they like humans consider?
How many places, for example, can a squirrel choose to deal with the problem of
hiding his nuts (a problem wh. may have been at the back of your mind) ?
There is not a logical set [or "space"] of places. There is a world of places
within a limited area (bounded by squirrels' travel capacity).
If the squirrel or you wish to find new hiding places within that area, there
are, first of all, not an identifiable set of locations. There are endless
places and always new places that you or he won't have thought about. And I
think it's a fairly safe bet that squirrels keep finding new ones.
Similarly if I give you the problem of finding places in an unfamiliar house to
hide nuts or jewels, you won't be able to approach it by consulting a logical
set/space of "hiding places" or "hiding places for nuts". And you won't be able
to google it, or find answers from any scientific or technological body of
knowledge. You have to get stuck in and start looking around.
Real world problems don't have identifiable sets of options. Scientific,
artistic, historical, technological, business, political, economic problems -
even physics problems - there aren't any neat sets of options for any of them.
That is the acknowledged challenge for modern logic and AI - they are
incapable of "real world intelligence" or "real world reasoning". Ben has just
written a book on it, and completely avoided the subject much as he did
creativity, by sticking to entirely toy logical problems.
Take any real world problem - "what were the causes of the French revolution?",
"how should Obama handle the election?", why did Obama lie:"yes we can"?, "what
shall we have for lunch?", "how can Sergio get a grant?".. and there are no
sets.
From: Sergio Pissanetzky
Sent: Tuesday, June 12, 2012 8:27 PM
To: AGI
Subject: RE: [agi] The 2 Tests of AGI - generalizability & creativity
Mike,
Civilized behavior, please Mike.
When a squirrel sees me coming, he climbs a tree. There is only one behavior in
response to that particular situation: how I was coming, how big I was, what
direction I was walking, where was the tree, how many branches it had, what
position he was in. You name it. There is a very complex, unique set of
circumstances surrounding the "see me coming" situation. But the response to
all that is only one behavior.
Next time, the circumstances may be almost identical, but the devil is in the
details: I said "almost." There may be the slightest of differences, yet the
squirrel reacts in a completely different way. Maybe the tree is a little bit
farther away, or was too thin, or the wind was blowing differently, who knows.
Yet, there is only one behavior that follows as a consequence of that set of
cirumstances.
The same may happen many times, and yes, it is a world of possibilites and a
world of behaviors. But for each combination of circumstances, the squirrel
executes only one behavior. And that behavior lasts for a short time only,
because the squirrel will adjust it again as soon as it encounters other
situations or risks.
Let's see if we agree up to this point. If we do, then I'll go one step
further.
Sergio
AGI | Archives | Modify Your Subscription
AGI | Archives | Modify Your Subscription
AGI | Archives | Modify Your Subscription
AGI | Archives | Modify Your Subscription
AGI | Archives | Modify Your Subscription
-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393
Modify Your Subscription:
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968
Powered by Listbox: http://www.listbox.com
