Before you use AI to grade this semester, read this

We are in the thick of grading season, and for many academics the temptation is obvious: paste a paragraph into ChatGPT, ask for a quick summary, generate draft feedback, and move on.

But here is the part many people have missed: under the EU AI Act, AI systems intended to evaluate learning outcomes are listed as high-risk. That does not mean every use of AI in teaching is banned, but it does mean assessment-related uses deserve much more caution than most institutions are currently communicating.

In other words, there is a real difference between using AI to help you think about a rubric and using AI to shape, score, or influence a student’s grade. The first may be relatively low risk; the second sits much closer to the area the Act is designed to regulate.

Here is a simple way to think about it:

• Generally lower-risk uses: brainstorming rubric language, drafting generic feedback phrasing, or helping structure teaching materials without using identifiable student work.

• Higher-risk territory: using AI to evaluate student submissions, generate grades or score recommendations, or materially influence assessment outcomes.

• Best practice: keep a human fully responsible for judgment, avoid uploading identifiable student data into public tools, and check your university’s AI policy before building AI into your grading workflow.

This is also where the ethical side matters just as much as the legal one. The European Commission’s guidance for educators emphasizes human agency, transparency, fairness, privacy, and accountability when AI is used in education. Those principles are especially important in grading, where students are affected by decisions that can shape progression, funding, and future opportunities.

So no, the answer is not “never use AI.” The better question is: what exactly is the AI doing, and does it influence assessment? That is the line more academics need to start taking seriously.

If you want to read the source documents yourself, start here:

• European Commission: Ethical guidelines on the use of AI and data in teaching and learning for educators

• EU AI Act Annex III: High-risk AI systems

A question for you: has your institution given staff any clear guidance on using AI in grading, peer review, or assessment design?

Enjoy your weekend - and hopefully you won't have too much grading ahead. 

All the best, 

Melanie