Skip to content
TheChatArchiveThe Chat Archive
GLMGLM-5 / Pony Alpha· unknownHugging Face Frontier Reasoning Traces

Build a course-level AI usage policy library: scenarios, guidance, and templates.

  1. User

    Build a course-level AI usage policy library: scenarios, guidance, and templates.

  2. GLM Assistant

    The user wants a "course-level AI usage policy library" consisting of scenarios, guidance, and templates. This is a practical, educational resource. The goal is to help instructors create policies for their specific courses regarding the use of Generative AI (like ChatGPT, Claude, etc.).

    Target Audience: Instructors, professors, instructional designers. Goal: To provide a structured, adaptable resource for setting expectations about AI use in the classroom.

    Structure of the Library:

    1. Introduction/Framework: Briefly explain why this matters and the spectrum of usage (from banned to encouraged).
    2. The Spectrum (The "Levels"): Define clear categories of policy (e.g., No AI, AI for Idea Generation, AI for Editing, Full AI Collaboration).
    3. Templates (The Policy Text): Ready-to-copy syllabus statements for each level.
    4. Scenarios (Application): Concrete examples of "Is this allowed?" for different levels.
    5. Guidance (Best Practices): How to talk to students, how to handle violations, and ethical considerations.
    6. Appendix/Resources: Citation guides, tools (optional but helpful).

    Drafting Content - Section by Section:

    • Framework: I need to move away from a binary "Cheating vs. Not Cheating" to a "Pedagogical Purpose" framework. Why do we want them to use (or not use) AI here?

      • Levels:
        1. Prohibited: Focus on foundational skills without assistance.
        2. Restricted (Assisted): Brainstorming/outline only, no drafting.
        3. Permitted (with attribution): Editing, coding assistance, but human creates the core.
        4. Encouraged (Integral): AI is a tool of the trade (like in marketing or coding courses).
    • Templates: These need to be written in "syllabus speak"—clear, authoritative, but supportive.

      • Template A (The "No AI" Approach): Needs to justify why (e.g., "We need to learn to write before we can edit").
      • Template B (The "Middle Ground"): Most common. "Use for ideas, cite if you use text."
      • Template C (The "Full Integration"): "Treat AI as a co-pilot. You are the pilot."
    • Scenarios: I need to cover different disciplines.

      • Writing: Generating text vs. grammar check.
      • Coding: Writing code vs. debugging.
      • Math: Solving the problem vs. checking work.
      • Group Work: Who is responsible if one uses AI?
    • Guidance/Best Practices:

      • Transparency: "Show your work" (prompt engineering logs).
      • Critical Thinking: AI hallucinates. Students must verify.
      • Data Privacy: Don't upload personal info.
    • Refining the Output:

      • Make it look like a "Library" or a "Toolkit." Use headers, bullet points, and clear separators.
      • Ensure the tone is balanced—not anti-AI, but pro-learning.

    Self-Correction during drafting:

    • Initial thought: Just list a few rules.
    • Correction: No, the prompt asks for a "Library." This implies structure and variety. I should organize it by "Policy Levels" because that's how instructors think (e.g., "How much AI do I want in my class?").
    • Critical addition: The "Rationale" section for each template is crucial. Instructors need to know when to pick which template.

    Final Polish:

    • Check for clarity.
    • Ensure "Scenarios" are mapped to the specific policies so the user sees the practical application.
    • Add a section on "How to Cite AI" since that's a common question.

    Let's build the response following this structure.

    • Title: The Course-Level AI Policy Library.
    • Section 1: The Spectrum of Usage (The Framework).
    • Section 2: Syllabus Templates (Copy/Paste).
    • Section 3: Scenario Analysis (Q&A).
    • Section 4: Operational Guidance (Citations & Logs).
    • Section 5: Conversation Starters.

    This looks solid. Proceeding to generate content based on this outline. This library is designed to help instructors move beyond simple "banned" or "allowed" policies toward a pedagogical approach that fits their specific learning goals.

    Below is a framework including Four Levels of Usage, Syllabus Templates, Scenario Guidance, and Student Contract templates.


    Part 1: The Spectrum of AI Usage (The Framework)

    Before selecting a policy, identify the skills you want students to demonstrate. If the skill is idea generation, AI is helpful. If the skill is writing syntax or calculation, AI may be detrimental.

    Level 1: The "AI-Inactive" Course (Restricted)

    • Philosophy: Focus is on foundational cognitive skills, mental models, and "unplugged" competencies that AI cannot replicate.
    • Use Case: Introductory language courses, hand-calculus, in-class essays, primary source analysis.
    • Rule: AI tools are not permitted for any part of the assignment generation.

    Level 2: The "AI-Assisted" Course (Structured)

    • Philosophy: AI is a study buddy or editor, not a creator. Students use AI for brainstorming, outlining, or debugging, but the final product must be human-generated.
    • Use Case: Advanced writing, coding, history research papers.
    • Rule: AI is allowed for specific stages (e.g., "idea phase only" or "grammar check only"), but the final text/code must be written by the student.

    Level 3: The "AI-Augmented" Course (Collaborative)

    • Philosophy: Students are evaluated on their ability to prompt, curate, and edit AI output. The student acts as a project manager.
    • Use Case: Marketing, business strategy, advanced coding projects, graphic design.
    • Rule: AI generation is allowed, but students must disclose the prompts used and explain the edits they made.

    Level 4: The "AI-Integrated" Course (Full Embrace)

    • Philosophy: The course teaches how to use AI as a professional tool. Assignments might require the creation of custom GPTs or complex workflows.
    • Use Case: Data science, prompt engineering courses, future-of-work seminars.
    • Rule: AI use is mandatory for assignment completion.

    Part 2: Syllabus Policy Templates (Copy & Adapt)

    Template A: The "Foundations First" Policy (Level 1)

    Policy on Generative AI: The goal of this course is to develop your independent critical thinking and foundational skills in [Subject]. Because AI tools (like ChatGPT, Claude, or Copilot) can bypass the struggle of learning these foundations, their use is not permitted on any graded assignments.

    Using these tools to generate ideas, outlines, or text constitutes academic dishonesty under the university’s Academic Integrity Policy. If you are struggling with the material, please utilize office hours and tutoring centers rather than AI tools.

    Template B: The "Thought Partner" Policy (Level 2)

    Policy on Generative AI: In this course, you may use Generative AI as a "thought partner" for brainstorming, outlining, or debugging code, similar to how you might use a tutor. However, the intellectual labor of writing and finalizing your work must be your own.

    Allowed: Asking AI to explain a concept, generate practice problems, or critique your draft. Prohibited: Asking AI to write your draft, solve the problem set, or translate passages for grading. Disclosure: You must include a statement at the end of assignments detailing what tools you used and for what purpose (e.g., "I used ChatGPT to brainstorm initial thesis statements").

    Template C: The "Co-Pilot" Policy (Level 3)

    Policy on Generative AI: This course prepares you for a professional world where AI is ubiquitous. You are encouraged to use AI tools to generate drafts, code, or visualizations. However, you are responsible for the accuracy and quality of the final submission.

    Requirements:

    1. Verification: You must fact-check all AI outputs. "The AI hallucinated" is not a valid excuse for incorrect facts.
    2. Documentation: You must submit an "AI Appendix" with every assignment. This appendix must contain:
      • The prompts you used.
      • A reflection on how you modified the AI's output and why.

    Part 3: Scenario Guidance (The "What If?" Library)

    Here is how to handle specific assignment types across different disciplines.

    Scenario 1: The Coding Assignment

    • The Risk: Students paste the prompt into AI and submit the resulting code without understanding it.
    • The Policy Strategy:
      • Low-Level Classes: Ban AI. Use oral exams or "code walkthroughs" where students must explain their code line-by-line to the instructor.
      • High-Level Classes: Allow AI, but grade heavily on code annotation. The grade is not "does it run?" but "explain why you chose this algorithm over another."

    Scenario 2: The Literature Essay

    • The Risk: AI produces generic, B-grade analysis instantly.
    • The Policy Strategy:
      • Focus on Specificity: Require students to quote from specific class discussions or physical textbooks that an AI wouldn't have access to (e.g., "Connect the theme of the reading to the specific debate we had in class on Tuesday").
      • Process Grading: Grade the outline, the rough draft, and the revision history, not just the final PDF.

    Scenario 3: The Discussion Board Post

    • The Risk: Generic, polite, and robotic responses that kill engagement.
    • The Policy Strategy:
      • The "Hot Take" Rule: Require students to take a specific, unpopular stance or relate the topic to a personal experience. AI struggles with genuine personal anecdotal connections.

    Scenario 4: The Data Analysis Project

    • The Risk: AI writes the Python/R script and interprets the graphs.
    • The Policy Strategy:
      • Level 3 Policy: Allow AI to write the code. The assessment focuses entirely on the Executive Summary. Can the student interpret the data better than the AI? Can they spot the nuance the AI missed?

    Part 4: Operational Templates

    The "AI Acknowledgement" Statement

    Require students to paste this at the top of every submission.

    AI Acknowledgement

    • I did not use any AI tools for this assignment.
    • I used AI tools for brainstorming/structure only. (Tool used: _________)
    • I used AI tools to generate content that I significantly edited. (Tool used: _________)
    • I used AI tools to generate content that I used verbatim. (Tool used: _________)

    Briefly describe how you used the tool: __________________________

    The "Prompt Log" (For Level 3 Courses)

    If you require students to show their work, ask them to attach a "Prompt Log."

    Prompt 1: "Explain the economic causes of the French Revolution focusing on bread prices." AI Output: [Paste relevant snippet] My Critique/Modification: The AI missed the impact of the American Revolution on the French debt. I added a paragraph specifically about war debts.


    Part 5: Student Guidance & Scripts (How to explain this to them)

    How to explain "Why?" to students

    Don't just say "it's cheating." Say:

    "I am not restricting AI to make your life harder. I am restricting it because this course is about training your brain. If you use a forklift, you won't get stronger muscles. If you use AI for this specific task, you won't build the neural pathways needed for the complex work you will do later."

    The "Hallucination Warning"

    Include this in your syllabus regardless of your policy level:

    The "Trust but Verify" Clause: Generative AI is prone to "hallucinating" (inventing fake facts, citations, or code libraries). If you submit work containing AI-generated fake citations, you are liable for academic misconduct regarding falsification. Ignorance of the tool's error rate is not a defense.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.