Here's a concise English summary of goshi_aoki's note on China's AI talent system (based on his Zhejiang University CS master's experience): China's AI edge stems from a hyper-competitive, merit-based pipeline where effort is heavily rewarded. - **Gaokao & high school**: 10M annual test-takers. Boarding schools grind 6am–11pm daily; no clubs/dating. Determines university/life path. - **University to grad school**: Low living costs (dorm ~¥2k/mo, cheap meals) let students focus. Kaoyan (master's exam) prep builds strong math/English foundations. - **Master's/PhD rigor**: Strict graduation — internal external reviews, defense, required publications/patents. PhD needs top-tier papers (e.g., CVPR/NeurIPS level). - **Zhejiang Univ example**: Intense research culture, long hours, performance pressure. Some view advisor guidance as harsh. - **Outcomes**: Top grads land high-salary roles (e.g., ¥15M starting at ByteDance). Continuous competition drives output; jobs demand results or risk termination. - **Vs Japan**: China repeats high-stakes exams (uni grad), turning university into extended competition rather than "break time." This capitalist-style selection fuels China's AI talent depth, though experiences vary. The system prioritizes volume elite performance.