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

Create a generative AI in teaching policy: use cases, offenses taxonomy, disclosures,…

  1. User

    Create a generative AI in teaching policy: use cases, offenses taxonomy, disclosures, and assessment redesign.

  2. GLM Assistant

    The user wants a comprehensive policy document for the use of Generative AI in teaching and learning environments. The policy needs to cover four specific areas:

    1. Use Cases: How can/should it be used?
    2. Offenses Taxonomy: What constitutes a violation?
    3. Disclosures: How must students/teachers declare AI use?
    4. Assessment Redesign: How should assignments change to account for AI?

    Target Audience: Educational institutions (universities, schools), faculty, and students. Tone: Professional, clear, forward-looking but cautious, pedagogical.

    Structure of the Policy:

    • Title: Clear and authoritative.
    • Preamble/Philosophy: Acknowledging AI is here to stay, focusing on integrity and learning outcomes.
    • Section 1: Permitted Use Cases (The "Green Light" and "Yellow Light").
      • Brainstorming, outlining, grammar checking, coding assistance, accessibility.
      • Different levels (Pre-K-12 vs. Higher Ed) might differ, but I'll aim for a general Higher Ed/High School framework as it's the most contentious area.
    • Section 2: Offenses Taxonomy (The "Red Light").
      • Need a graded approach. It's not all "expulsion."
      • Level 1: Minor citation errors.
      • Level 2: Significant unapproved use (polishing a whole essay).
      • Level 3: Full fabrication/contract cheating equivalent.
    • Section 3: Disclosures (Transparency).
      • How to cite? (IEEE, APA, or specific declaration form).
      • "I used ChatGPT for X."
    • Section 4: Assessment Redesign (The "Solution").
      • Move away from "write an essay on X."
      • Process-based assessment.
      • Oral exams/vivas.
      • In-class writing.
      • Multi-modal projects.
    • Conclusion: Summary.

    Drafting Content - Iteration 1 (Mental Outline):

    • Use Cases: Don't just ban it. Use it for ideation. Use it for explaining complex code.
    • Offenses: Plagiarism is the wrong word now. It's "Unauthorized Content Generation." Let's categorize by intent and magnitude.
    • Disclosures: Need a standard statement. "I pledge that..."
    • Assessment: If AI can do the homework, the homework is bad. Focus on critical thinking.

    Refining the "Offenses Taxonomy": Instead of just "Cheating," let's use terms like "Unauthorized Aid," "Data Fabrication," and "Intellectual Property negligence."

    Refining "Assessment Redesign": This is crucial. The policy needs to force teachers to change.

    • Old: Final essay.
    • New: Drafts + Reflection + Oral Defense.

    Formatting: Use markdown with clear headers, bullet points, and perhaps a tier system for offenses.

    Self-Correction during drafting: I need to make sure I don't sound too permissive or too restrictive. The "Human-in-the-loop" concept is vital. The policy must emphasize that learning happens with the human, the AI is just a tool.

    Let's build the document.

    Title: Institutional Policy for Generative AI in Education. Preamble: AI is a tool, not a replacement for thought.

    Section 1: Use Cases.

    • Tier 0: Prohibited (unless specified).
    • Tier 1: Support (brainstorming).
    • Tier 2: Integration (co-writing).
    • Wait, let's flip it to a "Spectrum of Use".
        1. Support: Accessibility, grammar.
        1. Exploration: Ideation.
        1. Integration: Co-creation (requires heavy disclosure).

    Section 2: Offenses Taxonomy.

    • Level 1 (Procedural): Failure to cite AI use when required.
    • Level 2 (Deceptive): Submitting AI output as one's own work without modification/effort.
    • Level 3 (Fabrication): Using AI to generate data, citations, or completing exams in real-time when prohibited.

    Section 3: Disclosures.

    • The "AI Declaration Statement."
    • Mandatory citation of prompts or tools used.

    Section 4: Assessment Redesign.

    • "Authentic Assessment."
    • Process over Product.
    • The "Viva" (oral defense).

    Final Polish: Ensure the language is policy-ready (e.g., "The Institution," "Faculty," "Student"). Add a disclaimer that this is a template and needs local adaptation.

    Let's generate the output based on this structure. Below is a comprehensive template for an Institutional Policy on Generative AI in Teaching and Learning. This policy is designed to be adaptable for universities, colleges, or K-12 environments. It addresses the four key pillars you requested: Use Cases, Offenses, Disclosures, and Assessment.


    Institutional Policy: Generative AI in Education

    Version: 1.0 Scope: All faculty, students, and staff involved in the learning process.

    Preamble

    This institution recognizes that Generative AI (GenAI) tools (e.g., ChatGPT, Claude, Midjourney, Copilot) represent a paradigm shift in information processing. Rather than banning these tools, we aim to cultivate AI Literacy, ensuring students learn to use these tools ethically, critically, and effectively. This policy outlines the boundaries of acceptable use, the consequences of violations, and the necessary adjustments to our assessment methods.


    1. Permitted Use Cases (The "Traffic Light" System)

    Because different disciplines require different skills, there is no "one size fits all" rule. Instructors will designate assignments into one of the following three categories:

    🟢 Green Light: AI Assisted (Encouraged)

    • Definition: GenAI is a collaborative partner. The student is the "pilot," and the AI is the "co-pilot."
    • Use Cases:
      • Brainstorming & Ideation: Generating outlines, mood boards, or lists of research topics.
      • Code Explanation: Asking AI to explain complex code snippets or debug errors.
      • Accessibility/ESL Support: Using AI to translate concepts, check grammar, or simplify dense reading materials.
      • Socratic Dialogue: "Roleplaying" with the AI to debate ideas or prepare for a discussion.

    🟡 Yellow Light: AI Supervised (Restricted)

    • Definition: GenAI use is permitted only for specific stages of the work, defined by the instructor.
    • Use Cases:
      • Editing: Using AI to refine a draft written entirely by the student (e.g., "Make this paragraph more concise").
      • Initial Research: Using AI to locate primary sources (which must then be manually verified).
      • Data Formatting: Using AI to clean up raw data sets or generate graphs based on student-provided data.

    🔴 Red Light: AI Prohibited (Human-Only)

    • Definition: The learning objective relies on the student performing the cognitive task without external digital assistance.
    • Use Cases:
      • Foundational Skills: Writing essays in a composition class, solving math problems without a calculator (if specified), or coding basics.
      • In-Class Exams: Closed-book assessments where memory and recall are tested.
      • Personal Reflections: Journals or reflection papers requiring personal voice and experience.

    2. Offenses Taxonomy (Levels of Violation)

    We move away from the binary term "plagiarism" toward a spectrum of "Unauthorized AI Assistance."

    Level 1: Procedural Violation (Negligence)

    • Description: The student used AI for minor assistance (e.g., grammar checking) but failed to disclose it as required by the assignment guidelines.
    • Detection: Often admitted by the student when asked, or inconsistent citation style.
    • Sanction: A warning and a mandatory revision of the assignment to include proper disclosures. Minor grade penalty (e.g., 5-10%).

    Level 2: Unauthorized Content Generation (Augmented Plagiarism)

    • Description: The student used AI to generate substantial portions of text, code, or imagery without permission and presented it as their own original work.
    • Detection: Significant deviation from student's previous voice, hallucinated citations, or AI detection software flags.
    • Sanction: Failure of the specific assignment. Mandatory workshop on AI ethics.

    Level 3: Contract Cheating / Evasion

    • Description: The student used AI to generate the entire assignment and utilized "humanizing" tools or specific prompts to evade detection. This constitutes a breach of academic integrity involving intent to deceive.
    • Detection: Oral defense failure (student cannot explain the work), specific inconsistencies in logic, or forensic metadata analysis.
    • Sanction: Failure of the course and referral to the Academic Integrity Board for potential suspension or expulsion.

    3. Mandatory Disclosures (Transparency Protocol)

    To ensure academic integrity, transparency is mandatory. If a student uses AI, they must document how it was used.

    The "Statement of AI Contribution"

    Students must append a statement to any assignment where AI was used (even in "Green" zones). This statement includes:

    1. Tool Used: (e.g., ChatGPT-4o, Gemini, DALL-E 3).
    2. Nature of Use: (e.g., "Brainstormed initial topics," "Checked grammar," "Generated Python code skeleton").
    3. Prompt & Response Log (Optional/Required): For higher-level work, the student may be required to submit a separate appendix containing the prompts they used and the raw output received.

    Citation Standard

    All AI-generated content that is included verbatim or paraphrased in the final work must be cited.

    • Format: [Tool Name], [Month Day, Year], Response to [Prompt Text].

    4. Assessment Redesign (Pedagogical Shift)

    To prevent violations and ensure learning, faculty must redesign assessments to be "AI-Resistant" or "AI-Inclusive."

    Strategy A: Process-Based Assessment

    • Concept: Value the journey of writing/creating, not just the final product.
    • Implementation:
      • Require submission of outlines, drafts, and revisions (track changes).
      • The "Viva" (Oral Defense): Randomly select students to explain their work verbally for 5 minutes. If they cannot explain what they wrote, the work is disqualified.

    Strategy B: The "Authentic" Assessment

    • Concept: Connect assignments to real-world, local, or current contexts that the AI is unlikely to know or handle well.
    • Implementation:
      • Ask students to apply a theory to a specific local event or a recent news article (less than 24 hours old).
      • Require analysis of data provided specifically by the professor (e.g., "Analyze this specific dataset," not "Find a dataset").

    Strategy C: AI-Inclusive Assignments

    • Concept: Force students to critique AI, acknowledging its limitations.
    • Implementation:
      • Prompt: "Ask ChatGPT to write an essay on [Topic]. Critique the AI's output for logical fallacies, hallucinations, and lack of depth. Rewrite the essay correcting these errors."
      • Prompt: "Use AI to generate code for [Problem]. Debug the three intentional errors in the code."

    Summary of Responsibilities

    • Faculty: Must clearly define the "Traffic Light" color for every assignment and redesign assessments to focus on critical thinking.
    • Students: Are responsible for understanding the restrictions for each assignment and strictly adhering to the Disclosure and Citation protocols.
    • Institution: Will provide support for AI literacy and detection tools, while ensuring due process for accused students.