Toward Embodied Superintelligence: An Integrated Architecture for Self-Regulating AGI

Read full text Summarise with VIRAT
Authors: Yuliia Yermolenko
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This work introduces a conceptual architecture for Artificial General Intelligence (AGI) based on modular cognition, temporal memory structuring, metacognitive regulation, and embedded ethical navigation. The proposed framework addresses key limitations of current AI systems, including the lack of structured memory, the conflation of reasoning and statistical association, the absence of internal self-regulation, and the externalized nature of ethical constraints. The architecture combines multiple specialized cognitive modules with metric-defined behavioral profiles, coordinated through a central synthesis mechanism and regulated by a non-participatory internal observer. Memory is organized into temporal layers (Present, Past, Future), enabling delayed consolidation and goal generation. Ethical alignment is implemented through a continuous Semantic Ethical Map (SEM), transforming ethics from a post-processing filter into a navigational field. Additionally, the system incorporates a physical decomposition module and a sensory–cognitive representation model based on dynamic curves, providing a foundation for embodied perception and grounding reasoning in real-world constraints. The architecture is designed as a self-regulating system with internal state, rather than a reactive tool. This publication represents an initial integrated formulation intended to establish conceptual priority and provide a foundation for further development and refinement.

Fields of Study & Subjects

Psychiatry Mental Health Neuroscience Cognitive Computing and Networks Ethics and Social Impacts of AI

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
21 March 2026
Language
EN