[ SYSTEM.LEVEL_03 / MY AGI ROAD ]

我的 AGI 之路My road to AGI

从 2022 年 ChatGPT 出现，到 2023 年预判 AGI 平民化，再到具身智能、工具链、Agent 与个人 AIOS 的连续实践。From ChatGPT in 2022, to a 2023 bet that AGI would become ordinary, and on to embodied intelligence, toolchains, agents, and a personal AI OS.
ROAD TO AGI / WHAT I ACTUALLY DID
2022
ChatGPT 横空出世ChatGPT arrives
惊闻 AI 涌现，重心从造酒到造物。The shock of emergence. The center of gravity moved from making wine to making things.
2023
AGI 平民化预判A public-AGI bet
历经波折，拥有第一个 ChatGPT 账号。After a long detour, the first ChatGPT account.
2024
打开心理学大门The door to psychology
43 岁报考清华心理学×AI 硕士，未录取；继续自学探索交叉方向。At 43, applied to Tsinghua's psychology × AI master's. Not admitted. Kept studying the intersection alone.
2025·上2025 H1
补足底层，视频 IPFoundations, and a video IP
具身智能、深度学习、强化学习、数学。Embodied intelligence, deep learning, reinforcement learning, mathematics.
2025·下2025 H2
Vibe Coding
VSCode、Cursor、Claude Code，开发产品。VS Code, Cursor, Claude Code. Shipping products.
2026·Q1
Agentic
OpenClaw、Hermes，Multi Agents。OpenClaw, Hermes, multi-agent systems.
2026·Q2
Architecture
Codex、Workbuddy、Paper Research、FDE，Harness。Codex, Workbuddy, paper research, forward-deployed work, harness.
SOURCE / 2023-05-27
Original note, 27 May 2023
MY WORKS / 作品与实践MY WORKS / PRACTICE
Gazebo simulation demo ▶

Gazebo 仿真 DemoGazebo simulation demo

搭建机器人交互测试平台A testbed for robot interaction

Reinforcement learning ▶

强化学习 (RL)Reinforcement learning

探索智能体的决策优化机制How an agent improves its decisions

Mathematics ▶

数学Mathematics

夯实理论基础，支撑算法创新The theory under the algorithms

Poster

Poster

葡萄酒海报生成Wine poster generator

Schedule

Schedule

AI 语音日程助手Voice schedule assistant

Paper

Paper

论文阅读与整理Reading and filing papers

Article

Article

微信公众号文章WeChat essays

AI 赋能培训：从认知到行动 · 不是 IT 技术课，而是认知升级与实操AI training: from understanding to action · Not an IT course. A shift in how people see, then what they do.

AI training for designers AI training, second scene
AI 5层发展Five layers of AI RAG & Wiki Ontology KnowledgeGraph Harness 私有化部署Private deployment