AI can help a lecturer prepare an outline, compare explanations or draft feedback. Its output still needs subject expertise, source checking and a clear teaching purpose. Start with a task where the proposed assistance can be evaluated, rather than assuming every use improves learning.
Establish How Your Students and Staff Use AI
Use local interviews, course feedback and a proportionate survey to understand actual practice. Usage differs by discipline, task and access to tools. A headline from a survey elsewhere does not establish what every student at your university does.
Publish clear course rules: permitted assistance, disclosure, citation, handling of personal data and how work will be assessed. Students and lecturers need a consistent explanation of the boundary for each task.
Useful Starting Points
- Preparation: draft a lesson outline from approved objectives and materials, then verify accuracy, depth and references.
- Assignment design: suggest variants and test them for equivalence, ambiguity and accessibility.
- Feedback: prepare a draft tied to an agreed rubric, with the lecturer checking the student’s work and final response.
- Course information: retrieve answers from approved materials with references and an escalation route when the sources do not support an answer.
- Review of difficulties: examine appropriately governed feedback or learning data, while avoiding unsupported inferences about individual students.
Human review is a practical control, but does not automatically remove AI Act high-risk obligations where the intended use falls within them.
Design Assessment for the Learning Objective
Decide what the work should demonstrate. Depending on the subject, an oral explanation, process record, draft comparison or supervised task may help assess understanding. No single format prevents all misuse. Avoid treating an AI detector’s output as conclusive evidence of misconduct.
Explain what students must disclose and what happens when there is uncertainty. Provide a fair review and appeal process, including for students who cannot access the same tools.
Control Data and Sources
Before uploading student work or teaching material, check permissions, the purpose and lawful basis for processing, service terms, retention and access. Consider confidentiality and intellectual-property rights as well as personal data. Hosting location alone does not establish compliance, and GDPR does not impose a blanket EU-only hosting rule.
Choose a model and deployment arrangement that suit the agreed data and operational requirements. Test unsupported answers, incorrect citations and instructions embedded in retrieved material. The system should not invent a source when it cannot answer.
Pilot and Evaluate
Compare the effort needed to prepare and verify materials, the quality of feedback and the effect on the intended learning task. Include error correction and staff training. Decide who owns the course materials, maintains sources and reviews changes to the tool.
GDPR and the AI Act at universities explains the legal questions to resolve alongside the educational design.
Developing a Teaching Workflow
Q-Dev can develop software for a defined university workflow. Integrations, access, deployment conditions, acceptance criteria, price and schedule are agreed for the project. An application that follows an audit belongs to the second stage of Q-Journey, including when AI is not part of that application.
Frequently Asked Questions
How many students use AI?
Usage depends on the institution, discipline and task. Use current research with its sample and methodology stated, and gather local evidence before making assumptions about your own students.
Can AI replace the lecturer’s judgement?
A teaching workflow should define where the lecturer checks and approves material. AI-generated feedback or assessment suggestions require verification and cannot be assumed accurate or appropriate.
How should a university regulate AI use?
Set clear rules for the learning task, permitted assistance, disclosure, data handling and assessment. Explain the review process and provide an accessible alternative where needed.
Is AI in teaching automatically lawful with human review?
No. The intended use, data processing, relevant transparency duties and any high-risk classification need assessment. Human review and EU hosting alone do not establish compliance.