[ExI] call for help

sam micheal micheals at msu.edu
Tue Jan 13 12:20:13 UTC 2009


Back in '95 when I did my paper research on AI and machine 
consciousness, I arrived at a model that does not differ extensively 
from the one in my recent paper on ACOMAS which can be found at:   
http://www.msu.edu/~micheal/acomas/

Now, after careful reconsideration, I arrive at the same two problems I 
faced at that time: specifications of 'visualization register' and 
'register feed'.

Consider the qualities of human capability of those things: we can 
dream; we can imagine; we have an emotional 'pool' which drives many of 
our thoughts and behaviors. But by design, ACOMAS will have no emotions 
nor emotional capability (at this point, I refuse to give that construct 
emotional capability in order to understand how much emotions drive us - 
and - in order to 'free' that construct from our emotional 'failings' 
(again, our predisposition to let our emotions 'drive' us and control 
our behaviors)). Night dreaming (as opposed to day dreams) is a 
wonderful illustration of our visualization capabilities. Part of our 
brains manufacture the dreams we experience. Another part experiences 
it. But it's as if we have a 3D projector in our minds that show us a 3D 
movie. This is an extraordinary capability not easily modeled or simulated.

The main reason it is difficult is because it is not easily amenable to 
conventional symbol conceptualization. The rest of the ACOMAS model is 
fairly amenable symbolically (because the components are symbol based) 
except for the senses .. So in this context, the senses and 
visualization register will have to be modeled and simulated on a 
completely different kind of technology (not based on conventional 
symbols). A virtual reality is not what I'm talking about (even though 
there may be similarities between VR and dreams). A "3D retina" is 
closer to what I'm asking for. Any objective specific constructive 
suggestions will be greatly appreciated.

Now back to 'register feed'. What I mean by this is the following. We, 
as human beings, have no problem to identify our motivations or why we 
"think a particular thought" - because we have feelings, desires, and 
dreams. ACOMAS will have no such benefits/attributes. Any feelings, 
desires, or dreams - we will "program" into ACOMAS. (How sad.) We could 
randomly feed its registers with random "thoughts" or "action plans" but 
that would not be very helpful to ACOMAS or the project. It would 
probably also invalidate its identity and confuse the hell out of it! 
(Imagine if someone had the power to fill your mind with random thoughts 
- it would probably drive you insane!)

So, what do we put in its registers? The prioritized goal list? That 
again defeats the purpose of the device. The purpose of ACOMAS is to see 
if "it can be done" (and how it might be done). If we give it this 
incredible capacity for original thought - yet put repetitive or 
unoriginal thoughts 'in its mind', we doom the device to be merely a 
plaything or curiosity. I contend both the visualization register and 
feed for registers can be creatively constructed so that machine 
capacity is not less than human capacity. (If I can scientifically 
imagine it, it can be built.)

So, if any of you dear readers have any objective specific constructive 
suggestions for visualization register or register feed (that is not 
emotion based), please don't hesitate to post on the forum or write to 
me directly.

Sam Micheal



More information about the extropy-chat mailing list