Your AI writing persona in one sentence

Recommended for: Courses with a writing or design assignment running over several weeks; start it in week one

Objective: Students can name their own boundary between judgement and assistance in one testable sentence, and can establish afterwards whether and why they crossed it.

Time needed: 15 minutes in week one, 10 minutes at submission · Group size: Any size; individual, with an optional anonymous plenary round

Background for the instructor

Courses set AI rules from above: this is allowed, that is not. That is necessary, but it leaves one thing untouched. A rule handed down is obeyed or evaded; a boundary the student wrote is tested. That is a different relationship to your own work.

Illingworth calls such a sentence a writing persona: not a description of style, but a statement of how you stand towards the tool. He writes it for professional writers; in teaching it works mainly because it stays visible while the writing happens, rather than only in the account given afterwards.

Mind what you are marking. A strict persona is not better than a permissive one. What counts is whether the sentence is precise and whether the student looks back at it honestly. A student who writes "I use no AI" and then does has learnt less than one who describes a wide boundary sharply and points at where they crossed it.

Procedure

1. Answer three questions, half a line each: which decision in this work will I not outsource under any circumstances? Where may a model take work off my hands? How would I notice that a text is no longer mine?
2. Compress the three answers into one sentence beginning "I use AI to…, but…". For example: "I use AI to test my structure and check my spelling, but I phrase every claim and every conclusion myself."
3. Put the sentence at the top of the working document, in the header, where it stays in view while writing. Not in a separate form nobody opens again.
4. At submission, ten minutes: does the sentence still hold? Note one place where you stretched or broke it, and what you were doing at the time.
5. Optional plenary: collect the sentences anonymously and show the group how differently the boundaries fall. Do not ask who wrote which.

Discussion questions

• Which decision turned out to be harder to keep than it looked when you wrote your sentence?
• Would your sentence also hold in a subject you feel less confident in? If not, what does that say?
• What struck you about other people’s sentences: was the boundary elsewhere, or only worded differently?
• Is there something you would have outsourced last year and would not now?

Assessment opportunity

The persona plus the look back at it is a compact, easily marked appendix. Mark precision and honesty, not strictness: "I used AI to rewrite my conclusion because I was stuck" is a better answer than a spotless persona with no friction in it. Where a course already asks for a transparency statement, this is its personal half - the statement says what happened, the persona says what the student intended.

Extension options

Have students rewrite the sentence at the end of the course without the old one in front of them, and only then compare. Shifts are more interesting than compliance. Across a programme the persona can travel between courses, so a series builds up rather than a one-off task.

Reference

• Based on Sam Illingworth, Keep Your Voice: A Slow AI Guide to Writing with AI and Staying Human (2026), theslowai.substack.com. The idea of a writing persona as a single sentence, kept visible during the work, comes from it.
• The guide is written for writers rather than teachers, and carries no open licence. This activity - its shape, its steps and its questions - is ours; what we took is the idea, not the wording.