MyTUA Portal All articles
Campus Technology & Tools

Reading the Room on AI: A Practical Matrix for Decoding Each Professor's Actual Rules

MyTUA Portal
Reading the Room on AI: A Practical Matrix for Decoding Each Professor's Actual Rules

Photo: National Aeronautics and Space Administration. Lyndon B. Johnson Space Center. 2/17/1973, Public domain, via Wikimedia Commons

Four years into the mainstream arrival of generative AI in coursework, the policies governing it remain stubbornly inconsistent. Two professors in the same department may issue opposite instructions. A single syllabus may prohibit AI in one paragraph and require it in another. Students who assume consistency tend to get burned, and students who assume the strictest reading everywhere often handicap themselves unnecessarily.

The practical answer is not to memorize rules. It is to build a small, repeatable system for reading them — one that produces a clear classification for every assignment and a record of what you did.

Three Dimensions, Not One

Most students think of AI policy as a single switch: allowed or banned. In practice, policies vary along three independent dimensions, and confusing them is the source of most violations.

The first dimension is prohibition: what the policy forbids outright. The second is disclosure: what the policy requires you to report even when the use itself is permitted. The third is assumption: what the policy silently expects you to know, such as the difference between editing your own prose and generating it.

A student can comply perfectly on the first dimension and fail the second. A student can disclose thoroughly under a policy that never asked for disclosure and create suspicion where none needed to exist. Separating the dimensions keeps you from solving the wrong problem.

Building the Matrix

Construct a simple table with assignments as rows and the three dimensions as columns. For each row, record the policy source — syllabus section, assignment sheet, or an explicit statement in class — and assign a short rating: prohibited, restricted, permitted with disclosure, or unrestricted.

Restricted is the category that does most of the work, and it is also the one students flatten most often. A restricted policy typically permits specific functions while prohibiting others: brainstorming and outlining allowed, drafting prohibited; grammar checking allowed, paraphrase generation prohibited; code debugging allowed, code generation prohibited. Write the specific permitted functions into the cell rather than a generic label. A year from now, you will not remember which functions were which.

When the Syllabus Is Silent

Silence is not permission, and it is not prohibition. It is an unanswered question, and it should be treated as a task rather than a risk to be gambled on.

Before assuming anything, check the three places policy usually hides: the assignment prompt itself, which sometimes carries a one-line restriction the syllabus omits; the course learning management system announcement feed, where instructors post corrections and clarifications; and the department's academic integrity page, which sometimes sets a floor below which individual instructors cannot go.

If all three are silent, ask. The question is worth asking early in the term rather than the night before a deadline.

How to Ask Without Sounding Like You Are Testing the Boundary

The framing that works is forward-looking and specific to the work in front of you. Describe the task, describe the tool use you are considering, and ask whether that falls inside the course's expectations. Vague inquiries invite vague answers; specific ones produce usable ones.

A workable formulation names the tool, the function, and the degree of use. For example: you are preparing a literature review, you intend to use a generative tool to locate possible sources that you will then read and verify yourself, and you want to confirm that this is acceptable. That is a question about the assignment. It is answerable. A question about whether AI is allowed in general is not.

Ask in writing when possible, and keep the reply. Email and learning management system messages create a record that protects both parties if the policy is reinterpreted later.

Documenting Use Without Turning It Into a Defense

Where disclosure is required, keep the record minimal and mechanical: the tool, the date, the function performed, and what you did with the output. A short appendix or a note at the end of a paper is usually sufficient. The purpose is transparency, not confession, and the tone should match the rest of the document.

Where disclosure is not required, a private log is still worth keeping. It costs almost nothing and resolves disputes quickly if a question arises about how a draft evolved. Store it with your course files rather than in a separate note-taking app you will forget.

Recheck at the Assignment Level

Course-wide policies set the default, but individual assignments frequently override it. A term paper may permit editing assistance while the in-class essay prohibits any tool use at all. Before submitting anything substantial, revisit the matrix row for that specific assignment and confirm it still matches the policy text. Policies also shift during a term, particularly after an integrity case in a large lecture. A row you filled in September may be out of date by November.

The system is not elaborate. It takes perhaps fifteen minutes to set up per course and two minutes per assignment to maintain. In exchange, it replaces guesswork with a documented, defensible reading of rules that are, for the moment, still being written in real time.

All Articles

Related Articles

LWSIS, Decoded: A Student's Guide to the Information System Behind Your Academic Record

LWSIS, Decoded: A Student's Guide to the Information System Behind Your Academic Record

Before You Click 'Buy': A Step-by-Step System for Cutting 2026 Course Material Costs in Half

Before You Click 'Buy': A Step-by-Step System for Cutting 2026 Course Material Costs in Half

When the Software Gets It Wrong: Decoding Your Degree Audit Before It Decodes You

When the Software Gets It Wrong: Decoding Your Degree Audit Before It Decodes You