What is the nature of fundamental requirements for one to conceive and develop General Artificial Intelligence (AGI)?

I would think, at the core, would be a modeling of the great adaptations of the actual human itself: its body , its senses, its responsiveness-es, its abilities, AND the abilities in cognition (representation and processing) that develop progressively and over EACH stage of ontogeny (largely: ages 0 - 18 yr. old) -- and with the latter influencing not only thought, but other responsiveness-es (e.g. the emotions; in other words, then: emotional development).

Now, you can start and stay with a good understanding of the human OR you can see the human as you see it and actualize your own hypothetico-deductive systems to have it progress in all relevant behavioral abilities.

Let's say you pick the first of these options (which I think is wise). Then what you need is a basic understanding of the human: human body, human senses, human responsiveness-es, and human abilities (INCLUDING those making "qualitative leaps" in their developments, during ontogeny -- these latter making up much of the "cognitive system", which can be conceived as the most-major AND central organizing system (as Piaget did) for all significant behavior patterns). Given all this, what else is needed? :

It should be clear that ALL faculties/basic abilities and responsiveness-es AND representation-and-thought abilities (including those "higher abilities": which "emerge" , unfold and develop with ontogeny) MUST BE GROUNDED CONCRETELY, specifically: clearly related to directly observable overt phenomena (behavior patterns). ALL OF IT.

Unless so-concretely "seen" (seen as at least related to key clear overt or overt-and-developing behavior patterns), it will not be possible to "mechanize" (here: program a machine) without being one with god-like insights. In other words, there will be NO AGI WITHOUT at least a basic (TESTABLE) understanding of ALL these fundamental behavior patterns (and their concrete "anchors", BOTH THOSE patterns continuing or now presently active, AND those key behaviors-and-circumstances active _AS_ the KEY POINTS of KEY DEVELOPMENTAL HAPPENINGS -- these creating new unfolding, lasting, and expanding representations and abilities). (These latter are also understood as clearly relating to some most-important directly observable concrete phenomena (behavior patterns, with corresponding situational aspects) and thus these also being "anchors" and, by virtue of some clear significant ongoing/continuing effects, they CONTINUE to be "anchors").

Now, does this mean the AGI developer needs no good thinking of just his own? NO. Unrefined inductive understandings (guesses) may be tested. And, proximate causal-type relationships can be hypothesized between THIS behavior pattern and THAT (even using some good hypothetico-deductive system, BUT this system must AT LEAST PRINCIPLED, IN TERMS OF LIFE (BIOLOGICAL) PRINCIPLES (e.g. a basic one: homeostatis)) [(I also suggest using the terminology of classical ethology, presented as-to-be-used in my earliest long paper.)]

The great news, of course, is: AGI People can test their overall system major-aspect-tried by major-aspect-tried over and over and thus be much facilitated in making corrections.

Now, what may be your final question: Where does one find such a wholly empirically-based, concrete-based understanding of behavior patterns/responses TO BEGIN WITH. Answer: I do my best to offer such a system through my "Developing a Usable Empirically-Based Outline of Human Behavior for FULL Artificial Intelligence (and for Psychology)" PROJECT and my " Human Ethology and Development (Ethogram Theory) : A Full-Fledged Paradigm for PSYCHOLOGY" PROJECT. And, I believe, that considered for being in the most-empirical terms and the most-concrete terms, the writings associated with these Projects are the best offered today. **

Start at my Profile, Brad Jesness and then look for those 2 just-named Projects (and see all Log Entries, aka Updates under them). Also see and read:

https://www.researchgate.net/publication/286920820_A_Human_Ethogram_Its_Scientific_Acceptability_and_Importance_now_NEW_because_new_technology_allows_investigation_of_the_hypotheses_an_early_MUST_READ

and

https://www.researchgate.net/publication/329428629_Essentially_all_Recent_Essays_on_Ethogram_Theory

and, when reading

https://www.researchgate.net/project/Human-Ethology-and-Development-Ethogram-Theory

, also see the Project Log of this Project to see many important Updates.

P.S. Plus, for a final 100 pages of recent essays (not among the 512 pages in the collection of recent essays, you already have been directed to), yet also very worthwhile essays composed after the 512 -page Collection , see them in this pdf: https://www.researchgate.net/publication/331907621_paradigmShiftFinalpdf

** FOOTNOTE: ** IMPORTANT NEWS ** : I recently presented summaries of my "system" that clearly indicate 2 types of basic and likely needed (and real) CONSTANTS: some constancy of our Memories faculties (on the more-purely endogenous "side") AND some constancy OF THE PROCESSES always involved in learning and development (most clearly and presently involving aspects of the external world), these simply based on the fact that the simple, well-defined FORMS of associative learning are intimately and always involved in all behavior pattern change (NOTE, here, that what is constant IS THE FORM, which otherwise differs enough in content to be seen (or have to be seen) as "different"). (Also NOTE: the constancies of the Memories seem also at least mostly a matter of "forms", though some clear abiding numerical limits (properly delimited) may [always] apply here and there (everywhere?) -- and, thank goodness for the latter: because we likely need somewhere always in the system some complete certainty, i.e. to some numerical degree.)

More Brad Jesness's questions See All
Similar questions and discussions