ON OPEN PROBLEMS IN ARTIFICIAL INTELLIGENCE: AN ONTOLOGY OF HISTORY AND CREATION

Le résumé fourni par la source

The question of what a historical knowledge base K ontologically carries — what processes over it constitute creation, and under what conditions a system can act autonomously without losing purpose — remains unarticulated in the existing literature. The Mesarović–Takahara general systems theory provides the formal structure of such a medium, but its ontological content is not specified: the framework does not formalize what history carries, how the creative process operates over it, or what structural conditions govern purposive autonomy. Existing approaches to formalizing creation in AI remain in the domain of attributes and observable patterns; this paper proceeds in the opposite direction — from the structural conditions of the possibility of creation. A minimal axiomatic foundation of the ontology of history and creation is introduced within the Mesarović–Takahara framework: history is defined as an abstract component that aggregates events into a chronologically ordered knowledge base K; the creative process is defined as non-deterministic recursion over K; the architecture of creation is defined as a recursive governing operator with two classes of binary relations. On this basis, the paper formalizes purpose as a target state within the domain of permissible reconfigurations over the binary model–component relation R, freedom as the absence of order constraints over R (constrained freedom as their imposition), and boundary ontological categories — love, eschatology, observability, and passage into a higher degree of existence — that determine whether a system remains in the creative regime or loses it. Verification is conducted through artistic research: a performance serves as a simulation model in which the proposed axiomatic framework is examined on a concrete scenic instance. The result is a formal framework in which history, creation, constrained freedom, and regulation become structural conditions of an evolutionary AI system — rather than its external additions. These definitions are introduced within the goal-seeking structure already present in the Mesarović–Takahara formalism, as the components its full operationalization requires; no additional postulates are needed beyond the formalism's existing vocabulary.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.