Sooner or later, every teacher meets the same situation: a well-written assignment arrives, and you wonder whether the student wrote it or an AI tool did. The instinct is to hunt for cheating. A more useful response is to change how assignments are set and assessed, so that learning is visible whether or not AI is used.
This guide gives faculty a practical, five-step approach that works in an ordinary college classroom, without expensive software.
Why blanket bans rarely work
Students can reach AI tools on their phones at any time, so a ban is almost impossible to enforce. It also pushes use out of sight, which means you lose the chance to teach students how to use AI honestly and carefully. Most colleges are moving toward clear rules about how AI may be used rather than pretending it does not exist.
Be equally careful with AI-detection software. Such tools can make mistakes, so treat a detector score as, at most, a reason to have a conversation — never as the only proof against a student.
Step 1: Write a simple class policy
Tell students, in writing, at the start of the course:
- when AI tools are allowed, for example for brainstorming or checking grammar
- when they are not allowed, for example in a closed-book test or a reflective essay
- how to disclose use, for example a two-line note: which tool, and what it was used for
- what happens if the policy is broken
Check first whether your college or university already has a policy, and keep yours consistent with it.
Step 2: Redesign assignments so thinking shows
Tasks that ask for a generic definition or a standard essay are the easiest to outsource. Assignments that need the student’s own context are harder to fake and more meaningful to do:
- Ask students to connect the topic to their own town, family business, internship or classroom experience.
- Use recent, local or class-specific material that a general tool is unlikely to know well.
- Ask for a decision and a justification, not only a description.
- Ask students to critique an AI-generated answer: what is correct, what is missing, what is wrong?
That last idea turns AI into a teaching tool. Students learn the subject and learn not to trust a confident answer blindly.
Step 3: Collect evidence of the process
Learning happens along the way, so assess the journey as well as the final file:
- an outline or mind map submitted first
- a rough draft with the student’s own notes
- a short reflection: “What was the hardest part? What did you change after feedback?”
- version history from the document, where the student is comfortable sharing it
Even a simple two-stage submission, outline first and final later, makes it much harder to hand in something the student does not understand.
Step 4: Add a short conversation
A three-minute oral check works better than any software. Ask the student to explain one argument in their own words or to answer one follow-up question. Students who did the thinking usually manage this easily; students who did not will find it difficult. Use it for a sample of students, or for submissions that concern you, and keep the tone respectful.
Step 5: Grade the skills you care about
Update your rubric so it rewards what AI cannot easily do for the student:
- clarity of the student’s own reasoning
- use of specific evidence and correct citations
- reflection and originality
- honest disclosure of the tools used
If your rubric rewards only polished language, you are rewarding what AI does best.
Teach students to use AI well
The aim is not only to catch misuse, but to build skill. Show students how to check facts, how to cite what they use and why they must never paste in text they cannot explain. Our guide for students, AI skills for commerce and arts students, covers the same habits from the student’s side and can be shared in class.
Get faculty ready together
Policies work best when the whole department agrees on them. A short faculty development programme helps teachers try the tools themselves, redesign one real assignment and agree on common rules. Our FDP on AI in Student Assignments is a five-day, in-person workshop at your college, built around exactly these steps. For teachers who want to go further, AI Unlocked — Data & AI for Educators & Trainers is a 10-week, no-code course. Browse everything for teachers on our faculty page, or request a proposal for your institution.
Conclusion
Banning AI or hunting for cheaters rarely solves the problem. What works is a clear class policy, assignments that need the student’s own thinking, evidence of the process, a short conversation when needed, and a rubric that rewards reasoning and honesty. Do these five things together, ideally as a whole department, and AI becomes part of your teaching instead of a threat to it.
A one-page checklist to take away
- Do we have a written AI policy for this course?
- Does each assignment need the student’s own context or judgement?
- Do we collect an outline, draft or reflection?
- Do we use a short oral check when needed?
- Does the rubric reward reasoning and honesty?
If you can answer “yes” to all five, AI stops being a threat to your assessment and starts being part of your teaching.
Frequently asked questions
Should colleges ban AI tools for assignments?
Blanket bans are very hard to enforce and hide AI use instead of teaching honest use. Most teachers do better with a clear written policy on when AI is allowed and how to disclose it.
Can AI-detection software prove a student used AI?
No. Such tools can make mistakes, so a detector score should be treated, at most, as a reason to talk to the student, never as the only proof.
What should a simple class AI policy include?
Say when AI is allowed, when it is not, how students must disclose it (for example, which tool and for what) and what happens if the policy is broken. Keep it consistent with your college's own rules.
How can I redesign an assignment quickly?
Ask for the student's own context, a decision with a justification, or a critique of an AI-generated answer. Collect an outline or draft first and add a short oral check for a sample of students.
Is there training for faculty on this?
Yes. Our FDP on AI in Student Assignments is a five-day, in-person workshop at your college, built around these steps.
Further reading and official resources
- UNESCO — Guidance for generative AI in education and research — a global starting point for policies on AI in classrooms
