Designing an Aligned Generative AI Course Policy: An Equitable and Transparent Learner-Centered Approach

Synthesis: An Aligned Generative AI Course Policy — a design case by McCorkle (2025) documenting how one instructor replaced a blanket GenAI prohibition with a policy whose allowed and unallowed uses are derived from what is specifically being assessed in the course. The method: challenge your own assumptions about generative AI, inventory every task a student performs on the project, ask of each task "what, specifically, am I assessing?", brainstorm an emerging GenAI workforce competency for that task, and then decide which concern takes priority — the need to assess student performance or the value of building the workforce competency. The result is a transparent, equitable policy that states its rationale to students task by task, and a secondary benefit the author did not expect: instructors are forced to think precisely about what they are actually assessing.

Overview

Institutional guidance on generative AI has lagged behind classroom reality, leaving instructors to write course-level policies themselves (Educational AI Policy). The author began the 2023–2024 academic year with a single prohibition-style policy across three courses, augmented by an offer students could invoke during office hours to co-create an amended version. No student ever took up the offer. Office-hour conversations triggered by observed use of teaching tools with built-in AI features (e.g., MagicSchool, Quizizz) revealed the real problem was not defiance but misalignment: students did not believe the policy applied to them because they did not see themselves as behaving dishonestly, and the instructor's assessment concerns — that the selected tools might substitute for authentic demonstration of the learning objectives — had never occurred to them. Each of these graduate students was a current or former K-12 teacher.

The design response was developed during a week-long Course Design Institute themed on equitable course design, drawing on transparent-design work on syllabi and rationale (Addy et al., 2021; Palmer et al., 2016; Winkelmes et al., 2019). The course was Multimedia Design, an advanced asynchronous online course in an Instructional Design & Technology program whose semester-long project is a project-based assessment: design and build a SCORM-compliant eLearning module in Articulate Storyline.

Positionality before policy

The design process opens with reflexive practice rather than rule-writing. The author describes the dread that followed ChatGPT's public release, an ambivalence between the workflow efficiencies GenAI offers instructional designers and the fear of grading AI-generated text, and a deliberate decision to move from suspicion to trust. McCabe's "20-60-20" heuristic on academic dishonesty — roughly 20% who will cheat regardless, 20% who never will, and the 60% in between who are the actual audience of any policy — is used to justify designing for that 60% (Pavela & McCabe, 1996). Students who are opportunistically dishonest, inadvertently plagiarizing, or simply unclear about instructor expectations (Lang, 2013; McCabe et al., 2012) can be nudged toward integrity by a policy that explains itself — a stance that positions Academic Integrity as a design problem of Trust and communication rather than one of enforcement.

Step 1 — Task inventory

For the multi-week project, the author conducted a simple task analysis (Jonassen et al., 1999), listing each step a student would take toward the finished product: topic selection, composing learning objectives, devising an assessment plan, prototyping assessment items, sequencing instruction, curating images and animations, developing slides, writing voice-over scripts, loading audio, creating captions, and publishing the SCORM package. This inventory was a private design document, never distributed to students.

Step 2 — Ask "what, specifically, am I assessing?"

The inventory was then mapped onto a working table with columns for (a) the task, (b) what exactly is being assessed and against which course learning objective, (c) an emerging GenAI workforce competency, and (d) the decision on whether AI is allowed. Instructional alignment — congruence among objectives, activities, and assessment (Smith & Ragan, 2005) — is applied here at task granularity rather than at course granularity. Tasks that are merely steps toward a product (rather than evidence of a learning objective) are marked as not assessed; the author notes that the exercise forced her to describe the performance she needed to observe far more precisely, and to consolidate and split tasks repeatedly. The process was explicitly non-linear.

Step 3 — Consider GenAI workforce competencies

Each task — including tasks that are assessed — was paired with a plausible professional use of GenAI by instructional designers, grounded in the literature on AI in instructional design workflows: brainstorming topics, prompting for objectives or lesson plans, generating images, generating slide decks, generating scripts, and text-to-speech or speech-to-text for accessible audio and captions. The design decision then weighs two goods against each other: the instructor's need to assess a capability versus the student's need to build an emerging workforce competency (Career Development and Readiness).

Step 4 — Decide, task by task

The resulting policy is deliberately uneven, which is the point:

• Allowed — topic selection and brainstorming. Learner analysis and needs assessment are assessed extensively elsewhere in the program, so GenAI use was permitted and the instructor's feedback moved to the quality and scope of the resulting idea.
• Unallowed — composing specific and measurable learning objectives. Even though prompting for objectives is a documented workforce competency, the author treats objective-writing as foundational and non-negotiable: a student who cannot compose objectives unaided cannot evaluate and revise an Large Language Models (LLMs)'s output for the same task. The stated priority is assessment need over workforce competency.
• Allowed — curating figures, images, and animations. The course assesses use of visuals per Mayer's Principles, not the ability to create an image or take a photograph; students must attribute generated assets as they would any other.
• Unallowed — developing slides with on-screen text and visuals. The task was split precisely so that image generation could be allowed while slide-level message design stayed with the student.
• Allowed — scripts, narration, and transcripts. Script-writing is covered extensively earlier in the program, and audio production and manual transcription are not course skills; students must still evaluate and correct AI output and document that process.

Step 5 — Write the rationale to students

A transparent policy for this design case is not a rule list but a rationale, composed as if speaking directly to students ("you will…", "we will…"), naming the Assessment that justifies each restriction — for example, that because the instructor is assessing the student's ability to compose specific and measurable objectives, GenAI assistance with that task is unallowed (Palmer et al., 2016; Weimer, 2013). Assignment types with a single clear rationale (discussion forums, design-case analysis, portfolio presentation) carry the policy in the syllabus plus a just-in-time reminder inside the assignment; the complex project carries a syllabus summary plus assignment-level call-outs. The author says she will keep the amendment-by-conversation option even though students never used it, because self-directed learners should be able to negotiate the policy against their own goals.

Implications

• Policy design is assessment design. Deriving permitted AI use from "what, specifically, am I assessing?" turns a compliance artifact into an alignment exercise that improves the course itself — and gives instructors a defensible answer when a student argues that a tool's use was not dishonest.
• Task-level granularity beats one rule per course. Different tasks inside the same project warrant different answers; a single blanket policy cannot express that, which is why task-type regulation is emerging as the practical pattern.
• Transparency is an equity move. Opaque policies assume shared background knowledge about authorship and attribution; making expectations and their rationale explicit dismantles part of the hidden curriculum and reduces the chance that policy failure becomes a disciplinary matter (Equity).
• Design for the persuadable majority. Framing the policy around Trust rather than surveillance targets the "60% in between," consistent with evidence that fear-based disclosure regimes push use underground (Reducing AI Misuse).
• Faculty development is the delivery mechanism. The approach emerged from a course design institute and spread by colleague curiosity — a model for Educational Development programs that want instructors to design their own policies rather than adopt template statements.
• The exercise is reusable at other levels. The author suggests teacher educators and program designers could apply the same inventory-and-prioritize method to program-level outcomes, and anticipates GenAI competencies migrating into formal learning objectives as the field's expectations stabilize.

Connected Concepts

• Educational AI Policy
• Academic Integrity
• Assessment
• Generative AI
• Curriculum Design
• Career Development and Readiness
• Framing AI Use for Students
• Educational Development
• Equity
• Higher Education
• Assessment Validity
• Authentic Assessment
• Trust
• Prompt Engineering
• Reducing AI Misuse
• AI Use and Disclosure Statements

Connected Articles

• A Guided Inquiry Approach to Students Co-Designing Generative AI Course Policies — Student co-design of GenAI course policy through guided inquiry (Hingle & Johri 2026)
• How Instructors Regulate AI in College: Evidence from 31,000 Course Syllabi — How instructors regulate AI across 31,000 syllabi; task-type differentiation (Chirikov 2026)
• A Comparative Analysis of Institutional and Course Generative AI Policies within Higher Education: Implications for Instruction in Computing Education — Institutional AI policy analysis in computing education
• What Does the Credential Still Certify? Cognitive Stewardship for AI-Mediated Education — Cognitive stewardship: delegation boundaries and evidence standards in AI assessment policy
• It's OK Because...": The Wild West of Student Rationalization of AI Use in Academic Writing — Five disconnect sites and 20+ student rationalizations of AI use
• Assuring quality learning in a gen AI-integrated future: The role of adaptive capabilities. TEQSA, June 2026 — Adaptive capabilities for assuring quality learning in a GenAI-integrated future
• Evaluation in the Age of AI: Output as Evidence of Learning — Evaluation in the age of AI: output as evidence and the disclosure trap
• A bit of chaos and madness: The AI Assessment Scale and the work of assessment reform — The AI Assessment Scale and assessment reform

Citation

McCorkle, S. (2025). Designing an Aligned Generative AI Course Policy: An Equitable and Transparent Learner-Centered Approach. International Journal of Designs for Learning, 16(2), 96–110.