From: Elizaphanian (Elizaphanian@members.v21.co.uk)
Date: Wed Oct 30 2002 - 08:02:18 GMT
Hi Patrick,
I'll come back about the specific Dennett point, but there's something I
need to clear up first. Probably as a result of my sometimes convoluted
manner of speaking, I think I have misled you on what my position is.
Let us distinguish 'human intelligence and reasoning' [humint] from 'general
intelligence and reasoning' [genint].
My position is that humint necessarily involves an emotional, ie a
physiological input. Specifically, that bodily responses to external stimuli
have cognitive content; human decision making is a function of refined
emotional reactions.
As a consequence, humint requires a human body (with all that that implies).
Genint I would describe as (something like) a patterned response to stimuli,
on the basis of algorithms. So a thermostat exhibits genint - when the
temperature goes up, it turns the heating off.
To my mind, AI research is concerned with establishing a very high quality
form of genint. No matter how high quality it becomes - and it could become
really very high quality indeed, I think the current 'expert systems' are
just the beginning - it will never replicate humint. (Unless, as I say, we
first reproduce the human life form, or its equivalent). [Although it is
conceivable that highly evolved genint could be a fourth level actor in a
way we don't fully understand. Are you familiar with the sci-fi author
Steven Baxter? His manifold sequence deals with some of this.]
I understand Penrose to be arguing that there are significant limits to how
far genint can go. Some of the most important problems cannot be rendered
algorithmically (thanks to Godel, amongst other things).
I don't think Penrose is making my point about humint, but I could be wrong
on that.
I think the most important element is that genint has no independent
capacity to make high quality decisions; that is, all decisions are
previously embedded in the programming. It is therefore unable to respond to
the new, to dynamic Quality. Which is why I don't think 'intellect' is a
good description of the fourth level, the standard account has no adequately
described 'choosing unit' analagous to the cell.
If we do achieve a very high quality genint actor - ie a robot, Data? - I
think it is essential that we follow Dr Asimov's wise guidance on their
construction.
Sam
www.elizaphanian.v-2-1.net/home.html
MOQ.ORG - http://www.moq.org
Mail Archive - http://alt.venus.co.uk/hypermail/moq_discuss/
MD Queries - horse@darkstar.uk.net
To unsubscribe from moq_discuss follow the instructions at:
http://www.moq.org/md/subscribe.html
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 10:38:07 GMT