暗信道与智能评价公式
2026년 05월 23일
ORIGINAL THOUGHT PAPER · 信息完整度框架 · 第一篇（共八篇）· V2 暗信道与智能评价公式 Dark Channel & Intelligence Evaluation Formula:A Unified Information Completeness Frame…
梁文锋160人科研团队 科研能力分析报告
2026년 05월 23일
RESEARCH REPORT · MAY 2026 · V5 梁文锋160人科研团队科研能力分析报告 邓巴数约束下的组织效率极限、AI编程工具市场攻击链、数据主权危机、七个边界条件与可观测触发指标 Liang Wenfeng’s 160-Person Research Team: Du…
Mythos模型架构和机制的 逆向分析
2026년 05월 23일
TECHNICAL ANALYSIS · MAY 2026 · V4 Mythos模型架构和机制的逆向分析 Reverse Engineering the Architecture and Mechanisms of Claude Mythos:Multi-Dimensional Technical…
AI实验的严谨性问题分析
2026년 05월 23일
CRITICAL ANALYSIS · MAY 2026 · V5 AI实验的严谨性问题分析 Rigor Problems in AI Experimentation:A Transmission-Variable Ablation Analysis of LLM Memory Consolidat…
AI本体论和方法论的博弈
2026년 05월 23일
ORIGINAL THOUGHT PAPER · MAY 2026 · V5 AI本体论和方法论的博弈 The Game Between AI’s Ontology and Methodology:Why Mathematics-Based AI Cannot Restore the O…
闭源生成式AI商业飞轮的 结构性减速
2026년 05월 23일
CRITICAL ANALYSIS · MAY 2026 闭源生成式AI商业飞轮的结构性减速 以验证瓶颈为主轴的五层递进分析含AI同行评审的叙事偏差实证检验 Structural Deceleration of the Closed-Source Generative AI Business Fly…
AI时代不确定性带来的人类健康危机
2026년 05월 23일
THOUGHT PAPER · MAY 2026 AI时代不确定性带来的人类健康危机 输入过载与调节失灵的双重危机：从进化失配到社会-认知负荷的系统动力学模型 The Dual Crisis of Input Overload and Regulatory Failure: A Systems Dy…
统一架构AI工厂设计方案
2026년 05월 19일
TECHNICAL DESIGN PROPOSAL · MAY 2026 · V8 统一架构AI工厂设计方案 面向长上下文Agent推理的大内存独立推理节点架构 Unified Architecture AI Factory: A Large-Memory Inference Node Archit…
RAG与AI的长期记忆技术
2026년 05월 19일
THOUGHT PAPER · MAY 2026 · V4 RAG与AI的长期记忆技术 从外部持久知识层到 Memory OS：大模型长期记忆的系统工程路径 From RAG to Memory OS: A Systems Engineering Path for Persistent AI Mem…
C端AI用户的贾维斯需求
2026년 05월 19일
ORIGINAL THOUGHT PAPER · MAY 2026 C端AI用户的贾维斯需求 从 OpenClaw 与 Hermes Agent 的兴衰看个人AI的结构性困境 The Jarvis Demand: Why Consumer AI FailsStructural Impossibili…