The Konrad Artificial
Consciousness Project: Technical Scope and Aims
Copyright Notice: This material was
written and published in Wales by Derek J. Smith (Chartered Engineer). It forms
part of a multifile e-learning resource, and subject only
to acknowledging Derek J. Smith's rights under international copyright law to
be identified as author may be freely downloaded and printed off in single
complete copies solely for the purposes of private study and/or review.
Commercial exploitation rights are reserved. The remote hyperlinks have been
selected for the academic appropriacy of their
contents; they were free of offensive and litigious content when selected, and
will be periodically checked to have remained so. Copyright
© 2011, High Tower Consultants Limited.
|
|
First published online 09:00 GMT 15th March 2011
1 - Introduction
Konrad is a "proof of concept" artificial
intelligence (AI) system. The first lines of code were written in August 2007,
and the first successful test run took place in March 2008 [see Press Release
(historic
pdf)]. Since then, development has been carried out in stages as further
areas of functionality have been added. This document provides some background
to the project and identifies some hopefully useful further reading.
2 - Project Background
The Konrad
software is the product of an academic collaboration between Derek J. Smith, Consultant Cognitive
Scientist, and International
Software Products, Toronto. It is grounded on the CA-IDMS
database platform and its ultimate aim is to validate the technical
hypothesis that the currency indicators maintained by CA-IDMS
have functional equivalents in the biological nervous system, where
"second messenger" neurotransmitter mechanisms [more on these]
mark the location of items of recently activated long-term memory. Currency
indicators - artificial or biological - are in this respect crucial to such
cognitive functions as perception, attention, phenomenal awareness, and behavioural planning and execution; any faculty, in fact,
where the mind needs to keep track of where it has got to in a prolonged
processing sequence. To the extent that this hypothesis might eventually be
supported it promises engineers more autonomous AI and robotic systems, and
psychologists greater understanding of the natural mind.
3 - Design Principles
Konrad is an "evidence-based" design, adopting and
doing its best to integrate a large number of individually well-respected
psychological theories, not least the following [in alphabetical order] .....
Arkin's (1990, 2009) theory of robotic autonomy
Arnheim's (1974) theory of perception in art
Austin's (1962) and Searle's (1969) theory of speech
acts
Brentano's (1874) notion of perceptual Vorstellung [=
presentation] [more
on this]
Craik's (1948) theory of "discontinuous" cyclical cognition [more on this]
Dennett's (1978) hierarchical model of cognition [more on this]
Edelman and Tononi's (2000)
"Dynamic Core Theory" of consciousness [more
on this]
Ellis and Young's (1988) "transcoding
model" [more
on this]
Freud's (1895) theory of physiological "cathexis" [see
Freud's Project]
Heidegger's (1927) notions of Dasein and In-der-Welt-Sein [more
on these]
Kernberg's (1967) theory of ego integrity versus "splitting" [more on
this]
Klein's "object relations" theory of
psychosexual development [more on this]
Lorenz and Tinbergen's "ethological" theory
of animal behaviour [more on this]
Pylyshyn's (1978) notion of "meta-representation" and "orders of
belief"
Rasmussen's (1983) hierarchical model of control error
[more on this]
Wertheimer's (1923) "Gestalt Laws" of perception
4 - Potential Application Areas
Konrad is
currently capable of simulating acts of cognition lasting up to 10 seconds or
so, and of then justifying its final response with a microscopically detailed hard copy telling you how it
did what it did and calculating what biochemical resources were consumed in
doing it.
Unlike most software, moreover, Konrad is designed to be as useful when it
fails as when it works. The system is
designed to provide fundamental theoretical insight into the structures of the
human self by recording how those structures can fail in characteristic ways.
For example, by manipulating Konrad's command parameters it is possible to simulate the
sort of incompletely connected, pointer-corrupt, experientially skewed, or
abnormally structured semantic networks which might underlie the following
.....
·
error in
general, and control error in particular
·
speech errors
and dysfluencies
·
learning
disorders such as dyslexia and dyscalculia
·
social
adjustment problems such as criminality, pathological gambling, oppositional
defiance, and elective homelessness
·
politico-affiliative problems such as political extremism and the
sort of suicidality seen in suicide bombers
·
psychiatric
disorders, especially those whose symptoms include hearing voices
·
post-traumatic
stress disorders, to the extent that the stressor in question involves damage
of some sort to the structures of the self
5 - Past Software Versions and Project Milestones
6 - Research Collaboration
CONTACT THE AUTHOR AT smithsrisca@btinternet.com
[Home]