Digital & AI English · Secondary 2 English · English Atlas
A student’s work can become cleaner, more fluent and more sophisticated while becoming less theirs. That is the central authorship risk of generative AI in English tuition.
The tutoring goal is not merely to produce impressive final text. It is to develop a learner who can generate ideas, make language choices, revise deliberately and stand behind the final answer. AI should therefore support the student’s process without quietly becoming the writer.
The authorship rule
Keep the core decisions with the learner: interpretation, idea selection, organisation, evidence choice, sentence production, revision judgement and final responsibility. AI may help by questioning, comparing, suggesting checks or generating practice material around those decisions.
1. Voice begins before wording
Student voice is not only style. It includes what the learner notices, which examples they choose, what position they take, what they consider important and how they qualify a claim. If AI chooses those things first, authorship has already shifted before any sentence is written.
2. Preserve the first idea generation
When idea generation is part of the learning goal, ask the learner to produce possibilities before AI suggests any. AI can later challenge the list, ask for missing perspectives or generate a counterposition. This keeps the student’s thinking visible.
3. Preserve the first plan
A ready-made AI plan can erase evidence about task interpretation and organisation. Ask the learner to create the first structure. AI may then propose a contrasting architecture for comparison, but the learner should decide what to keep and why.
4. Preserve original sentence production
If every difficult sentence is handed to AI, the student becomes dependent on external phrasing. Encourage the learner to write the sentence first, even imperfectly. AI can later offer alternatives or identify ambiguity, but the final sentence should be consciously chosen and understood by the learner.
5. Use AI as a critic, not a ghostwriter
Ask AI to identify questions, possible confusion, missing evidence or alternative viewpoints. Avoid prompts such as “Rewrite my essay to make it excellent” when the goal is to improve the student’s writing. Critique preserves the need for the learner to perform the revision.
6. Keep revision decisions visible
After AI feedback, require the student to mark each major change: accepted, rejected or modified. Add a brief reason. This creates a revision trail showing that the learner remained the decision-maker.
7. Do not equate AI polish with better voice
AI often shifts writing toward smooth, generic, balanced prose. That can remove specific personal detail, unusual but effective phrasing or the learner’s natural level of directness. Tutors should ask whether the edit improves communication or merely makes the writing sound more like a generic model.
8. Preserve age-appropriate language
A Secondary 2 student does not need to sound like a university essayist. Artificially elevated vocabulary can create register problems and obscure whether the learner actually controls the language. Precision is more valuable than adult-sounding complexity.
9. Preserve uncertainty when uncertainty is genuine
Student writing sometimes contains useful caution: “may”, “perhaps”, “in some cases”. AI rewrites can overstate. Tutors should protect justified qualification because it reflects disciplined thinking, not weak style.
10. Protect personal examples
When personal or observed examples are appropriate, they should come from the learner. AI-generated anecdotes can produce polished but hollow writing and may introduce fabricated experiences. The tutor can help the student develop a real example without replacing it.
11. Protect interpretation in reading responses
For text-based work, the student’s interpretation should appear before AI alternatives. AI may later offer a competing reading, but the learner should compare both against evidence. Authorship in comprehension means owning the reasoning, not merely copying the most convincing explanation.
12. Preserve spoken voice in oral practice
AI-generated oral scripts can sound unnatural and collapse under follow-up questions. Let the student answer naturally first. Use AI to ask harder questions, identify vague areas or suggest a different angle. Oral control should remain responsive rather than memorised.
13. Use version history as evidence
Keep the student’s first draft, AI-assisted notes, revised draft and final independent task where practical. The sequence reveals whether the learner is making increasingly better decisions or simply replacing their work with generated text.
14. Teach disclosure and task rules
Schools and assessment systems may set specific rules for AI use. Students should know when AI is permitted, when it must be disclosed and when it is prohibited. Ofqual’s 2026 resources emphasise that using AI to generate coursework without proper disclosure can constitute malpractice. Local school requirements should always take priority for submitted work.
15. Separate practice from submission
AI can be used more freely in a clearly labelled practice environment than in assessed work. A tutor may intentionally use AI-generated examples for critique, then require the learner to complete the actual school assignment under the school’s rules.
16. Maren: preserving analytical voice
Maren asks AI to improve a paragraph. The result sounds polished but replaces her specific interpretation with a generic claim. Her tutor returns to the original version and asks which insight she actually wants the reader to understand. Maren rewrites the paragraph herself, keeping the stronger structure but restoring her own reasoning.
17. Iona: preserving evidence judgement
AI offers Iona a confident inference. Instead of copying it, Iona compares it with her own reading and identifies that the AI version overstates motive. Her final answer keeps her narrower interpretation. Authorship is visible in the judgement.
18. Leonie: preserving execution ownership
AI creates an efficient revision schedule for Leonie, but it does not match her real timing failure. Leonie uses her own evidence to modify the plan. The tool can propose; the learner remains responsible for selecting the system that fits her performance data.
19. Use an authorship checkpoint
- Whose idea is this?
- Who chose the evidence?
- Who designed the structure?
- Who wrote the first version?
- Who decided which AI suggestions to accept?
- Can the learner explain every major change?
- Can the learner produce comparable work without AI?
20. Use a safe revision ladder
- Student writes first.
- AI asks questions or identifies possible issues.
- Student decides what is valid.
- Student revises personally.
- Tutor checks whether the principle is understood.
- Student completes a fresh AI-off task.
21. Where this page fits
This page owns student voice and authorship inside the Secondary 2 AI-assisted tutoring route. For AI job boundaries, use What Jobs Should AI Be Allowed to Do in English Tuition?. For checking AI advice, use How Should Students Check AI Feedback Before Acting on It?.
22. Further reading
- Ofqual: AI in coursework resources for schools
- OECD Digital Education Outlook 2026
- Singapore Student Learning Space: Guidance on Generative AI
23. Final test
Student voice is preserved when AI makes the learner’s thinking more visible and more deliberate—not when the tool quietly replaces it. The final work should remain something the student can explain, defend and reproduce at an appropriate level without the tool.
Explore the Secondary 2 English hub for related guides.