An Inquiry into the Homology of AGI Research Paradigms: A Cognitive-Philosophical Perspective

Contemporary research on Artificial General Intelligence (AGI) has fragmented into multiple competing research trajectories, each with distinct methodological commitments and technical implementations. The dominant academic discourse treats these lines of inquiry as mutually exclusive alternatives for constructing general intelligent systems. This paper argues that these seemingly conflicting paradigms are not rival solutions, but partial side-manifestations (partial bounded manifestations) of a unified intelligent ontology captured at different hierarchical cognitive layers. The fragmentation of current AGI research originates from a widespread methodological preference: engineering construction precedes ontological grounding. Drawing upon the methodology of scientific research programmes, this paper advocates an ontologically-first research stance from the perspective of philosophy of science and cognitive ontology. We analyze the inherent limits and implicit ontological presuppositions embedded within each existing AGI paradigm, contrast the “engineering-first” old framework with an “ontology-first” new framework, and discuss the implication that raw computational power alone is insufficient for the emergence of stable general cognition. This article remains at the level of methodological and philosophical analysis; no concrete algorithmic, architectural or engineering implementation details are presented.