In classic Husserlian phenomenology, the self is the referent of a grammatical placeholder for the narrator of a viewpoint. (E.g. in "I see the tree.". The thing denoted by "I" is the self).
Since all symbolic operations proceed by pattern-recognition (I have a proof, if anyone cares), the self is therefore what recognizes patterns, establishing a symbolic narrative between an object (picked out of the noise), and a viewpoint.
If this theory is right, then intelligence should be the process of establishing reusable scripts from a spontaneous narrative caused by successive pattern recognition.
AI is therefore a trivial scripting algorithm driven by 2nd-order pattern-recognition. i.e. recognizing patterns in scripts, as well as IRL, and then selecting scripts whose early conditions match the current states of affairs, and then sorting scripts for the best-payoff, and picking the best-paying script, and then performing the narrator's part.
And yes, it can be wrong.
Ray Van De Walker rgvandewalker@juno.com