Derek wrote:
>>Building computer systems that generate pictures or videos as their way of
>>communicating with us could be a very lucrative addition to computer
>>applications that include cognitive models of their users (instead of
>>focusing solely on generating natural language), because most of us do
>>process visual information so well.
-------------------------------------------------------------------------------------------
1.
Derek,
A simple movie, like a visual slide show or cartoon, would be a fairly straight
forward implementation for certain symbolic/connectionist and other models.
The architecture I work on is set to do that as a medium term goal. Any
representation system (language) is learned within the contextual patterns its
used, so a visual language (shapes), motion language (ultimately gestures),
text, sound (un-implemented), etc. of all sensor type can be integrated with
the core system as interchangeable patterns.*
You can simply read off the state from any region to see what it's thinking
within specific and multimodal sensors, as well as what it's attention is on
and the focus level of that atttention; e.g. looking at the colorful red and
blue car in the parking lot, but narrowly focus on the rust on the door handle
or a wide defocus on the entire parking lot and the heat from the pavement -
all in the same situation. When this situation is recalled or referenced, the
scene is pulled-up and it will generate its version of what it was focused on
(rust on the door handle or heat). It can then shift it's focus outward or to
other regions in the scene, modify a feature (say change the red colors to blue
on the car etc).
A movie from its output would look like the jerky camera movements and quick
focus change in a show like "NYPD Blue" or the new "Battlestar Galactic" space
scenes or History Channel's "Dogfight" sky scenes.
*pattens (for the model) have spatial and serial components to varying degrees
for each sensor: e.g. bodyspace maps easy to vision, sound to change (motion
sequence), force (touch) to vision,...for the sensors that don't map well, just
force it and you get your less seemless metaphors : a sweet sound, but left the
room with a bitter taste
2.
Derek wrote:
>>....because most of us do process visual information so well.
It makes it easier to program visually to see what the system is doingas a
whole or by states
--- On Wed, 11/26/08, Derek Zahn <[EMAIL PROTECTED]> wrote:
From: Derek Zahn <[EMAIL PROTECTED]>
Subject: RE: [agi] The Future of AGI
To: agi@v2.listbox.com
Date: Wednesday, November 26, 2008, 11:02 AM
#yiv2133712726 .hmmessage P
{
margin:0px;padding:0px;}
#yiv2133712726 {
font-size:10pt;font-family:Verdana;}
Although a lot of AI-type research focuses on natural language interfaces
between computer systems and their human users, computers have the ability to
create visual images (which people can't do in real-time beyond gestures and
facial expressions). Building computer systems that generate pictures or
videos as their way of communicating with us could be a very lucrative addition
to computer applications that include cognitive models of their users (instead
of focusing solely on generating natural language), because most of us do
process visual information so well.
This is really narrow AI I suppose, though it's kind of on the borderline. It
does seem like one of the ways to commercialize incremental progress toward AGI.
Derek Zahn
supermodelling.net
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/
Modify Your Subscription:
https://www.listbox.com/member/?member_id=8660244&id_secret=120640061-aded06
Powered by Listbox: http://www.listbox.com
