How to frame AI at the university without cheating?

Imagine a university teacher preparing a didactics course. She uses AI (artificial intelligence) to find examples, rephrase an instruction, and save two hours in her week. In class, her students then ask her: "Are we allowed to use it for the assignment?" No one has the same answer.

At another institution, management wants to modernize assessments with digital tools. The idea seems appealing, but one question blocks everything: what student data will be collected, stored, or reused? The issue is no longer just about allowing or prohibiting. It needs to be framed.

What is it, concretely?

Generative AI refers to tools capable of producing text, plans, exercises, or answers based on a written request. ChatGPT or Copilot are well-known examples. In pedagogy, these tools can help a teacher prepare materials, vary instructions, create a practical case, or propose remediation paths. The mechanism remains simple from the user's side: a request is made, the tool responds, then the human checks, adapts, and decides. The key point is there. If the tool replaces the student's effort or the teacher's judgment, the evaluation loses its meaning. Well-framed, AI can remain a useful assistant without becoming the pilot of the course.

Concrete case: what to do and what not to do

In Vancouver, the University of British Columbia (UBC), a major Canadian university, supports its teachers in the face of the massive arrival of generative AI in courses, assignments, and pedagogical preparation. The problem is very concrete: some teachers want to use it to save time, while others fear that students will submit work they do not really understand.

The temptation would be to allow everything or to prohibit everything. UBC adopts a more cautious approach: clarifying expectations, rethinking assessments, and checking privacy. The university notably states that an AI tool cannot be imposed on students without prior analysis of privacy or without an acceptable alternative.

What can be done

Draft a usage rule for AI for each course.
Explain the link between the rule and the learning objective.
Request intermediate steps: plan, draft, justification, oral.
Train teachers and students on errors, biases, and privacy.
Analyze the collected data before making a tool mandatory.

What not to do

Impose a tool without examining the requested personal data.
Confuse speed of production with pedagogical quality.
Delegate grading to a machine without human oversight.
Publish a vague rule, without examples by type of assignment.
Assume that all students already know how to use these tools correctly.

Questions to Ask Before Acting

Is the use of AI allowed, prohibited, or framed in this specific course?
Must the student declare when they used it, and for which part of the work?
Does the requested tool collect a name, address, copy, or learning traces?
Does the assessment still measure the student's real understanding?
Is there an alternative for students who refuse or cannot use the tool?
Do the instructions distinguish between brainstorming, language correction, planning, and complete writing?
Does the institution have a common policy, or does each teacher improvise alone?

UNIVGA Viewpoint

For the School of Education & Pedagogy, the real issue is not the trend of technological effects. It is the quality of the training. In Gabon, as in many Francophone African countries, purchasing a tool before training teachers would be a mistake. Pedagogical scripting, the construction of authentic assessments, and student support must come before equipment.

The issue of student data also deserves particular vigilance. Where do the copies, names, written productions, and exchange histories go? The APDPVP (Gabonese authority responsible for the protection of personal data and privacy) reminds us that the processing of personal data may be subject to declaration or authorization depending on the case. A responsible university cannot therefore request any tool without examining this point.

UNIVGA advocates a simple position: AI can support the teacher, but it must not reduce human supervision. It can help prepare, vary, simulate, and remedy. It should not solely determine the value of learning. For future educational engineers, school administrators, and trainers, the expected competence will be to design clear rules adapted to local realities: connection costs, access to paid tools, languages of instruction, and learner needs.

How to frame AI at the university without cheating?

What is it, concretely?

Concrete case: what to do and what not to do

Questions to Ask Before Acting

UNIVGA Viewpoint

Sources

Data centers: how to secure the entrance?

Digital cadastre: how to avoid land errors?

Can AI enhance the reliability of accident investigations?

Digital cadastre: how to avoid land errors?

Data centers: how to secure the entrance?

License Plate Cameras: How to Monitor Without Excess?

Leave a reply Cancel reply

Professional Certificate in Fisheries Transformation and Conservation Techniques — Transformation of Fisheries Products

Shopping cart