📄 中文摘要
人工通用智能（AGI）被视为人工智能领域的圣杯，各大科技公司正在投入前所未有的资源来追求这一目标。然而，目前尚无统一的正式定义，仅存在一些经验性的AGI基准框架。该研究旨在构建一个通用的、代数的和范畴论的框架，以描述、比较和分析不同的AGI架构。通过这种范畴论的形式化，可以比较多种候选AGI架构，如强化学习（RL）、通用人工智能（Universal AI）、主动推理（Active Inference）、对比学习（CRL）、基于模式的学习等。这将有助于明确揭示它们的共性和差异。
工作论文：朝着一个范畴论的人工通用智能比较框架迈进
📄 English Summary
Working Paper: Towards a Category-theoretic Comparative Framework for Artificial General Intelligence
The pursuit of Artificial General Intelligence (AGI) has become a paramount goal in the AI field, with major tech companies investing unprecedented resources. However, a unified formal definition of AGI is lacking, and only a few empirical benchmarking frameworks exist. This study proposes a general, algebraic, and category-theoretic framework for describing, comparing, and analyzing various AGI architectures. Such a category-theoretic formalization enables the comparison of different candidate AGI architectures, including Reinforcement Learning (RL), Universal AI, Active Inference, Contrastive Learning (CRL), and Schema-based Learning. This framework aims to unambiguously expose the commonalities and differences among these architectures.