Mechanism-Based AGI Architectures

Amar Nath Thakur

Abstract


ABSTRACT Artificial General Intelligence (AGI) aims to develop machines that can perform any intellectual task that a human can do. Most present AI systems are taskspecific and rely heavily on data-driven learning without deep understanding of internal mechanisms. Recently, researchers are focusing on mechanismbased architectures where intelligence emerges from structured cognitive components such as memory, reasoning, perception, planning, and learning modules that interact in systematic way. These architectures attempt to mimic how human cognition works rather than only optimizing statistical performance. This paper reviews mechanism-based AGI architectures, their design principles, cognitive inspirations, and implementation strategies. Various architectural models like symbolic-subsymbolic hybrids, cognitive architectures, neurosymbolic models, and world-model based agents are discussed. Advantages, challenges, and future research directions are also presented. Tables and figures are included to show architectural comparisons and module interactions.

KEYWORDS: Artificial General Intelligence, Cognitive Architecture, Mechanism-based Design, Neurosymbolic AI, World Models, Memory Systems, Reasoning Engine.


Full Text:

PDF 122-133

Refbacks

  • There are currently no refbacks.