[ExI] Conscious AI or a Zombie?

[Stuart LaForge]

Quoting Jason Resch via extropy-chat <extropy-chat at lists.extropy.org>:

> If you believe that philosophical zombies are logically impossible, then
> consciousness is logically necessary (in the presence of certain
> behavioral, sensory, and introspective capacities).

Actually, I would more rigorously say that philosophical zombies are  
impossible precisely BECAUSE they violate what I call causal awareness  
principle or CAP. Causal awareness is hereby defined as the ability to  
sense, act on, and react to ones environment as well as ones own  
effect on it. Anything that is causally aware is able to act  
rationally. Consciousness is necessary, but not sufficient, for causal  
awareness, and therefore, rational action. However, something cannot  
act rationally without being conscious too. Acting in ones own best  
interests requires consciousness. A sentient being must be conscious  
of those interests and the means to achieve them in the context of a  
changing environment.

Here rational action is used as a proxy for intelligent behavior. That  
is to say that philosophical zombies cannot exist because they would  
have no way to display intelligent goal-seeking behavior because they  
would not be conscious of any goal; Therefore, they could not be  
conscious of how to navigate the environment to achieve a goal; nor,  
would they be conscious of whether they had achieved the goal or not.  
That is to say, that logically, a philosophical zombie does not have  
the logical properties necessary to fit its own definition.  
Philosophical zombies are an unwarranted assumption to make the Hard  
Problem of consciousness seem like a different question than the  
so-called Easy Problem, whereas, CAP says that the complete solution  
of the Easy Problem, is also the solution to Hard Problem.

While pre-programmed behavior could mimic rational action in the short  
term, any sufficient change in the p-zombie's environment, like a new  
obstacle, would thwart it, and expose it as a zombie. In the harsh  
world of nature, philosophical zombies, even if they came to exist by  
some extraordinary chance, would quickly go extinct.

Therefore philosophical zombies, as opposed to fungal zombies, are  
both logically and physically impossible. It is, in short, impossible  
to do what humans evolved from worms to be able to do, without being  
in some measure, more conscious than a worm.

> Accordingly, I believe
> that consciousness was an unavoidable consequence during the course of
> evolution of life on Earth as once nature created creatures having certain
> capacities/abilities, consciousness had no other choice but to exist, it
> became logically necessary.

Yes, there is certainly a natural history of consciousness. You can  
look at the evolution of the nervous system of vertebrates through the  
process of encephalization and decussation and see a causal narrative.  
The moment life decided to detach itself from the ocean bottom and  
move around looking for food, a complete, or open, digestive system  
with a seperate mouth and anus became advantageous. Once it started  
moving mouth first through the world, sensory organs like eyes or  
antennae became advantageous; Moreover, those creatures which evolved  
sensory organs like taste-buds and eyes near their mouths, as opposed  
to near their anus, had a survival advantage.

Once an organism had a concentration of senses on its rostral or front  
end, that end became differentiated from the caudal or rear end of the  
organism. The cluster of sensory organs became a rudimentary head. As  
part of that differentiation called encephalization, it became  
advantageous for that organism to develop an organ to process the  
sensory information and locate it near the sense organs. The organ I  
speak of began as a small nerve cluster. As successive generations of  
the organism started moving through the environment faster, sensing  
and avoiding danger, finding food and resources, weathering natural  
disasters and extinction events, and generally leading more  
complicated lives, it finally evolved into the conscious brain.

> The same, I believe, is true for AI. It is unavoidable when we create
> machines of certain abilities, and I believe existing software is already
> conscious. For example, my open source project on artificial sentience
> could be such an example: https://github.com/jasonkresch/bots

Yes, I agree. These AI learned how to play soccer/football on their  
own. They are certainly conscious of one another, the ball, and their  
goals and that consciousness allows some very complex goal-seeking  
behavior to emerge. By the CAP, these AI agents in the following video  
presentation of a recent Science paper by Lui et al.

https://www.youtube.com/watch?v=KHMwq9pv7mg&t=10s.

Some might object because the consciousness is expressed within a  
virtual setting. But that's all right because Google built bodies for  
the little guys:

https://www.youtube.com/watch?v=RbyQcCT6890

If you push one over, they stand right back up. So yeah, the CAP says  
they are rudimentarily conscious because they display causal awareness  
and rational action. They lie somewhere between thermostats and humans  
on the consciousness scale. Someday, their consciousness may far  
surpass ours.

Stuart LaForge