Kotia: Toward a Civilization of Intelligence Beyond the Human--AI Divide

Contemporary artificial intelligence development is strongly organized around increasing the capability of individual models. Human intelligence, however, is historically maintained through language, education, tools, institutions, specialization, social learning, external memory, cumulative culture, and intergenerational transfer. Treating ``human intelligence'' and ``artificial intelligence'' as separate primary categories can therefore obscure a more general question: what allows intelligence to develop, remain socially usable, and survive beyond particular subjects? This paper proposes "Kotia", a term for an intelligent subject considered prior to classification by biological or artificial implementation. Kotia does not assert biological, cognitive, phenomenological, legal, or moral equivalence among implementations. Its intended novelty is methodological: it changes the unit and order of analysis. The accompanying "Kotia substitution method" first formulates a claim at the Kotia level, then substitutes particular implementations, and only afterward identifies implementation-specific mechanisms and constraints. The framework then moves the principal continuity criterion from the current intelligent subject to the civilization that preserves socially usable intelligence across changing subjects. The central question is therefore: "What changes in AI research when the unit of intelligence is moved from the individual to civilization?" The framework rejects universal individual capability as the default ideal of intelligence. A Kotia may remain bounded, specialized, dependent on other Kotia, and unable to perform many tasks. Operational intelligence is instead treated as a coupled relation between subjects and a persistent "intelligence environment". Memory, knowledge, computation, skills, procedures, evaluation, permissions, and tools can partly exist as independently maintainable "intelligence resources". Resource use is bidirectional: outcomes and failures can return to Kotia as experience, while aggregated experience can improve the resources. Intelligence may also be transferred partially across Kotia through education, records, apprenticeship, explanation, or other implementation-appropriate mechanisms. These elements motivate a "civilization of intelligence": a system in which socially usable intelligence can persist, accumulate, be taught, evaluated, governed, challenged, selectively forgotten, and inherited across turnover of individual Kotia, technical generations, and providers. The paper therefore treats cross-subject continuity of socially usable intelligence as a central design criterion and proposes a replacement-retention research test for measuring how much validated institutional capability survives subject replacement. The resulting research program shifts attention from model capability alone to the joint development of intelligent subjects and intelligence infrastructure. In some domains, the limiting factor may be the maturity of shared memory, qualification, resource quality, interoperability, governance, or succession rather than further base-model capability. This "intelligence-infrastructure bottleneck hypothesis" is posed as an empirical question, not as a claim that model development has ceased to matter.