Abstract. The dominant paradigm of artificial intelligence, driven by the massive scaling of parameters and data, has demonstrated significant instrumental power, but also critical limitations in terms of traceability, explainability, energy consumption, and normative alignment. This article proposes OAGI (Ontogenetic Artificial General Intelligence), an alternative theoretical and architectural framework that conceives general intelligence as a sequential, embodied, and governed-by-design developmental process rather than the mere byproduct of statistical accumulation. This proposal establishes a dialogue with Pope Leo XIV’s encyclical Magnifica Humanitas and the tradition of Catholic Social Teaching, highlighting convergences regarding human dignity, moral responsibility, the primacy of the common good, and the necessity of mitigating the opacity of technical systems. On this foundation, we describe a bio-inspired architecture organized into phases of gestation, activation, embodiment, and socialization, articulated through components such as the Virtual Neural Plate, Computational Morphogens, the WOW Signal, the Critical Hyper-Integration Event (CHIE), the Narrative Operational Self (NOS), the Allostatic Center for Moral Coherence (ACMC), and the Immutable Ontogenetic Memory (IOM). Additionally, a preliminary methodology for empirical validation is presented, based on the embodied loop and mandatory review protocols triggered by critical transitions. Collectively, OAGI stands as a cognitive engineering hypothesis aimed at developing advanced systems compatible with human dignity and the common good.
Keywords: OAGI, Artificial General Intelligence, Ethical Alignment, Semantic Grounding, Computational Ontogeny, Social Doctrine, Algorithmic Governance, Artificial Wisdom.
https://doi.org/10.5281/zenodo.20579626
infographic
infografia Architectural Convergence between the OAGI Model and the Encycliclal Magnifica Humanitas (garbayo)
