Principles of Artificial Intelligence (Peking University): Review

Who’s writing this

I’m Lin Zhou — a Chinese full-stack developer who moved into AI applications, writing here as a bilingual reviewer. I didn’t study through this course myself; I went through it because a relative prepping for China’s graduate entrance exams asked which Chinese course could serve as a classical-AI backbone. My own classical foundations were patched together piecemeal, and I’ve always envied people who got to walk the AIMA line systematically. In the Chinese-speaking world, the closest thing to that is Peking University’s Principles of Artificial Intelligence, taught by Professor Wang Wenmin.

What it is

The course was designated a national-quality online open course by China’s Ministry of Education in 2018, and that accreditation holds up. The design is bilingual: Chinese instruction, with slides, assignments, and quizzes in both Chinese and English — genuinely useful if you’re building terminology across languages. Multiple universities list it as a designated online resource in their syllabi. Watching and quizzing is free; the verified certificate costs an undisclosed fee that isn’t visible without logging in — check at enrollment.

The status needs spelling out. The most recent run on Chinese University MOOC (icourse163) ran March–June 2024 and is marked closed, with no evidence of 2025–2026 sessions there. The same course is currently running on XuetangX — July 2026 to January 2027, with about 279,000 cumulative enrollments on the mirror page. It hasn’t been discontinued; its center of gravity has partly shifted platforms. If you want the cohort experience with deadlines and a forum, confirm the XuetangX window first.

Teaching language is Chinese. For English-only readers, that settles the question early — see the alternatives.

What’s inside

Twelve weeks, twelve lectures, roughly 40–60 hours total, following the AIMA arc: AI overview and history, intelligent agents, uninformed and heuristic search (A*), local search and evolutionary algorithms, adversarial search (alpha-beta pruning, Monte Carlo methods), constraint satisfaction, knowledge representation and logical reasoning, Bayesian networks, classical and real-world planning, and a four-lecture machine learning block covering supervised, unsupervised, and reinforcement learning. Search, reasoning, planning, learning in one line — a structural rigor rare among Chinese MOOCs. Assessment is transparent: 30% unit tests, 30% assignments, 40% final; 60 passes, 85 earns an honors certificate signed by the instructor.

Where it falls short

The syllabus is classical-first: post-2020 deep learning and LLMs are barely there. Dedicated public discussion of the course has also become scarce, a sign of declining community activity.

Chinese University MOOC dropped free e-certificates in October 2019, and paid certificates are widely seen as low-value. More important, take the prerequisites seriously — data structures, algorithms, probability, linear algebra, ideally some ML background. True beginners will hit the wall.

Alternatives

• CS50’s Introduction to AI with Python — similar classical coverage with heavier, more open-ended projects, but English-taught and more demanding.
• Andrew Ng’s ML Specialization — go straight to machine learning; PKU wins on the full search-reasoning-planning arc.
• Hung-yi Lee’s NTU lectures — the opposite bet: frontier generative AI versus classical foundations. Choose by goal.

Bottom line

If you want a complete, rigorous map of classical AI in Chinese, this is still the first choice — free, bilingual, nationally accredited. But it’s a classical course, not a frontier one, and the icourse163 sessions have contracted: confirm the XuetangX enrollment window before planning around it. For English-only readers, CS50 AI gives you the same classical arc without the language barrier.