AI Agent Guidelines for CS336 at Stanford

This document provides guidelines for using AI agents (specifically Claude) in Stanford's CS336 course. It establishes best practices for leveraging AI assistance while maintaining academic integrity and learning objectives. The guidelines cover appropriate use cases, limitations to understand, and expectations for student work.

Key Points

• •Use AI agents as learning tools to understand concepts, not as shortcuts to bypass learning
• •Verify all AI-generated code and explanations independently before using them
• •Disclose AI assistance in your work and document which parts received AI help
• •Understand AI limitations: hallucinations, outdated information, and reasoning errors are common
• •Use AI for brainstorming, debugging, and explaining concepts, but write core logic yourself
• •Ask clarifying questions to AI agents to deepen understanding rather than accepting answers at face value
• •Be aware that AI may generate plausible-sounding but incorrect solutions
• •Use AI to explore alternative approaches and learn multiple ways to solve problems
• •Document your AI interactions as part of your learning process
• •Maintain academic integrity by ensuring submitted work reflects your own understanding

Workflow Diagram

Concepts