www.Conscious-Robots.com - Re:Pensabot Test

The Latest
13th Finnish Artificial Intelligence Conference (STeP 2008) We will create conscious machines in the future Searching for Spirit RoboChamps Competition Microsoft Robotics Developer Studio 2008 CTP April Higher Order Theories of Consciousness Feelix Growing Project

Popular
PSYCHE-D Discussion Forum (Theoretical emphasis) I-Cybie Review A Robot in Every Home MobileRobots Pioneer 3 (P3 DX) Cognitive Robotics Machine Consciousness and Strong AI Robotic Bugs

Conscious-Robots.com Forum

Boardwalk :: Forum List

Machine Consciousness Forums

Test for Consciousness

<< Start < Prev 1 2 Next > End >>

Pensabot

User

Junior Boarder

Posts: 8

Karma: 2

Re:Pensabot Test - 2008/02/20 10:10 You are right.

Main test procedure wants a "machine learning process" on high-level concepts (spontaneous model-building) emerged from its focus memory, then reasoning and behavior (behavior to explore the linked mirrorimages with motion feedback)

* external mirrorimage ---> internal visual action ---> context prediction ---> internal mirrorimage ---> intention and action.

"Memory flows" means a pattern copy by neuron activation intensity at real-time of the stimuli reaction between brain regions. For example, neocortex (perception, high-level thought,...) and hippocampus (memory formation, episodic,...)

3-axis representation of learning space (like learning curve) helps to watch the zoom-out machine pattern signals deployed on human brain model functional neuroimaging, but humanoid-like mind architecture depending on anatomy: from autonomous mobile robots (like Pioneer) to humanoid robots (like Honda ASIMO). I'm especially interested in this last type of robots.

The goal of this cybernetic test is to evaluate its android level of consciousness, by model matching between relative measurements of high-level concepts representation of its temporal memory about itself.

Post edited by: Pensabot, at: 2008/02/20 10:32 Chris.

Blog http://pensabot.com

The administrator has disabled public write access. Please, register to participate in the forum.

Raúl

Moderator

Posts: 325

Karma: 4

Re:Pensabot Test - 2008/02/22 12:47 I think I understand your point. The problem I see is that even though you have a humanoid robot like Asimo, that doesn’t necessarily means that it has a control architecture inspired in the human brain (the neuron-engineering approach you actually mentioned already in this forum). Don’t you think other sorts of tests are needed to evaluate the level of consciousness of robots that are not designed following a human-brain inspired model? Raúl Arrabales Moreno. conscious-robots.com

The administrator has disabled public write access. Please, register to participate in the forum.

Pensabot

User

Junior Boarder

Posts: 8

Karma: 2

Re:Pensabot Test - 2008/03/03 13:31 Yes, I do.

Maybe, it is possible to classify and join our tests by two groups: (1) Introspection tests; (2) evidence by psychoanalysis.

(1) measures of internal memory and scientific method;

Mind-architectures like human brain model are in development phase, so introspection tests is a parallel process to evaluate next years depending on maturity level.

Current strength of ASIMO is motion balance and a few of human interaction, consciousness level is 0, but ASIMO body is ready to modelling its mind like a human brain model (basic version)

I think other distinct mind-architectures are, for example, a lot of autonomous organisms groupped in a neural network that may reach levels of consciousness too (robots networks). An introspective test by analysis memory is possible if we know the mind-architecture, H-T-M pathways and to be able to access its memory.

(2) evaluation depending on external observations and interpretations.

Also, I think is possible to experiment with discrete level of consciousness with a sort of non-introspective tests or psichoanalysis by interaction with an observer of behavior.

Mensaje editado por: Pensabot, el: 2008/03/03 13:43 Chris.

Blog http://pensabot.com

The administrator has disabled public write access. Please, register to participate in the forum.

Raúl

Moderator

Posts: 325

Karma: 4

Re:Pensabot Test - 2008/03/07 11:01 Yup, I am actually working on the definition of a set of behavioral tests to assess the level of consciousness of artificial agents. The work is still incipient, but I think is a possible way to go in order to obtain a mature and reliable test for machine consciousness.

Establishing a scale or ordered set of levels of consciousness is also a quite complex task, and it is not free of controversy. Anyhow, I aim to define a set of well characterized levels of (machine) consciousness and then define the necessary behavioral tests for each level.

The distinction between state consciousness and creature consciousness is another variable to take into account in this scenario. One have to take into account that the same subject can present different states of consciousness during time… Raúl Arrabales Moreno. conscious-robots.com

The administrator has disabled public write access. Please, register to participate in the forum.

<< Start < Prev 1 2 Next > End >>

Boardwalk :: Forum List Machine Consciousness Forums Test for Consciousness