[4L4-GS-1a] Fundamental AI, theory algorithm

1:30 PM - 1:45 PM JST(4:30 AM - 4:45 AM UTC)

[4L4-GS-1a-01] An AGI Architecture with Hemispheric Integration of Monad-Modules

〇MASARU HIRAKATA¹, Chong MA¹, Rafal RZEPKA², Tomoaki YAMADA³, Kiwamu KASE⁴ (1. National Maritime Research Institute, 2. Hokkaido University, 3. NIPPON KAIJI KYOKAI, 4. RIKEN Center for Advanced Photonics)

Keywords:

Artificial General Intelligence (AGI),intelligence development,neural network architecture

This study proposes AGI architecture based on hemispheric integration of monad-modules, designed to unify right-hemispheric generative differentiation and left-hemispheric symbolic stabilization. While LLM lacks right-hemispheric processes such as continuous fields, which are essential for creativity, autonomy, and action grounded in relationships with others.The proposed architecture adopts monads with differential dynamics as its basic units, integrating modules corresponding to the right hemisphere, the left hemisphere, and the corpus callosum. Furthermore, a developmental-stage model is incorporated to connect sensory inputs (vision and speech) with concept formation through hemispheric interference, enabling meta-cognitive control and naming behaviors. As an initial testbed, the study addresses the gavagai problem to examine sensorimotor category differentiation.This research provides a foundational framework for integrating neural, philosophical, and computational perspectives on emergent intelligence. At the same time, it proposes a structural perspective on how the safety and ethics of socially deployed intelligent systems can be structurally ensured by internalizing otherness, using the maritime domain as a representative example of advanced AI-supported social infrastructure.