AI Agent Guidelines for CS336 at Stanford

github.com
June 1, 2026
3 min read
67/100
Summary
AI coding assistants like ChatGPT and Claude Code are designed to support students in CS336 by providing explanations, guidance, and feedback without completing assignments for them. The course emphasizes substantial Python/PyTorch coding with minimal scaffolding to enhance the learning experience.
Key Takeaways
- AI coding assistants in CS336 are designed to help students learn by providing guidance, explanations, and feedback rather than completing assignments for them.
- Students are expected to write substantial Python/PyTorch code with limited scaffolding, and AI assistance should enhance their learning experience.
- AI agents should ask clarifying questions and reference course materials instead of giving direct answers to student queries.
- The focus of AI assistance should be on suggesting improvements, debugging strategies, and understanding concepts rather than providing complete solutions or implementing code directly.
Coverage & discussions
Hacker News · github.com
Community discussions
Community Sentiment
Positives
- The guidelines for AI agents in education acknowledge the reality of AI use, aiming to integrate these tools into learning rather than banning them outright, which could enhance student engagement.
- This approach encourages students to explore the constructive use of AI agents as teaching tools, potentially fostering a deeper understanding of their applications in real-world scenarios.
Concerns
- The enforcement of AI agent guidelines appears unenforceable, raising concerns about their effectiveness in promoting academic integrity among students who may bypass them.
- There is skepticism about whether these guidelines can genuinely improve learning outcomes, as students might still rely on AI tools without truly engaging with the material.
Related Articles

10-202: Introduction to Modern AI (CMU)
Mar 1, 2026

CS336: Language Modeling from Scratch
Jun 1, 2026

AI makes the easy part easier and the hard part harder
Feb 8, 2026

Expertise in the age of AI
May 29, 2026

Atlassian’s Mike Cannon-Brookes: AI agents need a name tag—and a human to be accountable for their actions
Oct 19, 2026