独立研究计划，聚焦"受治理的人工智能自主性"的哲学与工程问题，同时涉及软件工程、信息论与数学领域。
最后更新：2026 年 7 月
AI 治理、控制与自主性
核心研究方向：真正"治理"一个 AI Agent 意味着什么——区别于仅仅"指令"它或"对齐"它。
| 标题 | 状态 | 链接 |
|---|---|---|
| Intelligence Beyond Knowledge: Control, Architecture, and the Structural Law of Artificial Agency | 已接收，AI and Ethics（Springer），2026 年 7 月 | Zenodo · SSRN · PhilPapers |
| From Knowledge-Centric Computation to Control-Centric Intelligence | 预印本（独立分支，未进一步发展） | Authorea |
| Intent-Conditioned Computation: A Systems Framework for Foundation-Model-Based Intelligent Systems | 预印本；审稿中 | SSRN · Authorea · PhilArchive |
| Probabilistic Execution Beyond Classical Systems: A Context-Dependent Coalgebraic Framework with Canonical Minimal Realisation | 预印本 | Authorea · HAL · SSRN |
| PPES / P-FAC：Probabilistic Program Execution Semantics | 预印本；预审批中 | Authorea |
| Governed Multi-Agent Dynamics: Stability, Stratification, and Complexity Bounds under Endogenous Constraint Fields | 预印本（ANANKE-1） | Zenodo |
| A Formal Architecture for Governed Enterprise AI Agents: Forms, Lifecycle, and Accountability | 预印本；审稿中（ANANKE-2） | Zenodo |
| Modeling Governance Constraints in Heterogeneous Multi-Agent Systems: A Dynamical Framework | 预印本（UAD-1A） | Zenodo |
| UAD / UAST（完整标题待确认） | 预印本 | Zenodo · Authorea |
| Intelligence Beyond Knowledge: The PAK Framework（PAK / "Prompt≠Knowledge"） | 预印本 | PhilPapers · Authorea |
| Architecture as Law: The Structural Necessity of the Fifth Condition for Artificial Intelligence | 预印本 | Zenodo |
面向 LLM／Agent 时代的软件工程
| 标题 | 状态 | 链接 |
|---|---|---|
| Rethinking Software Engineering Foundations in the Large Language Model Era（PRISM-1） | 预印本 | Zenodo · SSRN |
| PACT: A Reference Viewpoint Taxonomy for Software-Intensive Systems | 预印本 | Preprints.org · HAL · SSRN · Zenodo |
| AOP，变体 A（完整标题待确认——属于 Agent Operational Protocols 系列） | 预印本；审稿中 | Authorea · SSRN · Zenodo |
| AOP，变体 B（完整标题待确认，为变体 A 的配套论文） | 预印本 | Authorea · SSRN · Zenodo |
| AVDM: Architecture Viewpoint Decision Model（内部另有 "Architectonic Value-Driven Methodology" 的说法，两者命名尚未统一） | 预印本；审稿中 | TechRxiv · Zenodo |
| AADM: An Architecture Artifact Decision Model for Risk-Aware Documentation Scope | 预印本 | TechRxiv · Zenodo |
| ARCM（完整标题待确认） | 预印本 | Authorea · TechRxiv |
| RADAR（完整标题待确认） | 预印本 | Zenodo |
信息论、经济学与社会
| 标题 | 状态 | 链接 |
|---|---|---|
| Expansion-Compression Loop | 预印本 | Zenodo |
| Semantic Entropy and Structural Invariance | 预印本 | Zenodo |
| Mean Reversion or Innovation Collapse? | 预印本 | Zenodo |
| Signalling Inflation and Rational Adaptation | 预印本 | Zenodo |
| Cyber-Grooming — AI-Mediated Phatic Communion | 预印本 | Zenodo |
| Civilizational Control Theory（CCT） | 预印本 | PhilPapers · SSRN |
| The Structure and Dynamics of a Future Hybrid Society of Humans, Robots, and AI Agents: A Multidisciplinary Perspective with AI Distinctions | 预印本 | PhilPapers · HAL · SSRN · Zenodo |
| TAP：语言的物理学（Physics of Language） | 预印本 | PhilPapers · Authorea · SSRN |
| TAT：AI 三难困境（The AI Trilemma） | 预印本 | PhilArchive |
| The Inference Rent: Manufactured Scarcity, Physical Constraint, and the Durability of Platform Rent in the Age of Foundation Models | 预印本 | SSRN |
数学
| 标题 | 状态 | 链接 |
|---|---|---|
| A Trial Potential on the Logarithmic Lattice: Definitions and Elementary Properties | 预印本 | Zenodo |
| Additive Sieve in the Logarithmic Lattice: A Geometric Reading of the Multiplicative Structure of ℕ | 预印本 | Zenodo |
正式发表（同行评审）
| 标题 | 出处 |
|---|---|
| Innovative Research on IoT Architecture and Robotic Operating Platforms: Applications of Large Language Models and Generative AI | IEEE Xplore（另有 TechRxiv 与 HAL 镜像） |
完整引用格式请见随附的 BibTeX 文件。欢迎指正与交流。