I would highly recommend we research these details by observing and evaluating the social behavior of a similarly social species (i.e. doing the discovery and analysis with some species other than humans).

----------------------------------------------------------------------------------------------------------------------

[ If something develops fast, yet it is complex (upon analysis), then this neither means that "it's really simple" (which simply would be false) NOR does it indicate the 'great power' of some mythological [pure] learning (because you like to think of 'learning' like that). What it does indicate is clearly some innate guidance is involved (and reminds us again that perhaps, indeed, innate guidance and learning OCCUR SIMULTANEOUSLY with all the most significant sets of learning during ontogeny). ]

Dear Andrew Silletti

Knowledge is a construct (a useful one), reflected in and defined by behavior. 

If you look into my perspective, you would see I am totally an empiricist and more of a behaviorist than B.F. Skinner (or any of that lot) -- there was room for a LOT of improvement to any old-time behaviorism (including social learning 'theory').

P.S. Andrew, how is it that you discover and see behavior?  In the case of one giving a response such as yours, this is a very important big question(s). "Behavior" should not be just intuited or what you want it to be like (but without more real perspective, that is what it will be).

[ The only organism on which old-time 'reinforcements' work best is those who follow the party line of the old-time behaviorists and simply believe what they believe -- though it is, in actuality, very ill-defined.  (I was under them as a student for many years). ]

Dear Brad, I can put my definition of AI (I): Collective Type (swarm, social) - may be it will help some):

The Definition of AI (I): Collective Intelligence.

«The Collective Intelligence can occur in the group of AI-robots with AI of Individual Type, and try to reach the following objectives: (1) to enlarge the population by all controlled means: reproduction, protection of members of the population from attacks on life and health, increase life expectancy and living conditions of members of the population, аnd (2) to organize mutual activity of members to enlarge the living recourses for population».

For the occurrence of the phenomenon of the AI Collective Type must be simultaneous fulfillment of the following conditions:

1)      the group of AI Individual Type Robots should exists;

2)      the presence of an inherited / learned kinship characters and concept of “one rank” or hierarchical one;

3)      the presence of an inherited / learned concept of group interest.

4)      the presence of an inherited / learned “process  / plan” to reach the objectives on 3).

5)      the presence of an inherited / learned «incentive / penalty» (may be several types) to enforce members of the group to follow the Group Priorities».

The author is sure that all functions mentioned above can be modelled with known approaches but the real models are out of the represented article.

Dear Sergey Korneev

Your items sound very promising for AI.  Many of the constructs must be discovered as they are for human science -- we psychology people have to get away from models we come up with and reasoning by analogy. 

Some AI models (perhaps yours) seem better than the models of many psychologists in some ways (e.g. no pre-judging the place or nature of 'learning').

Regards, Brad

Dear Brad, thank you, unfortunately I am just busy to read thoroughly your articles - I will do it a little later. I wanted to enlarge a little your imagination) - may be it will be in the service. Best regards, Sergii