The genuinely good uses
Dismissing AI outright misses real value, because used as a study aid it does things that genuinely help learning:
- Explaining a stuck concept another way. When a textbook explanation hasn't landed, asking for it "explained more simply" or with a different analogy can unstick a student — a patient, on-demand re-explainer.
- Generating practice questions. "Give me five exam-style questions on price elasticity" turns AI into a practice-question machine — and practice and retrieval are exactly what the evidence favours.
- Acting as a quiz partner. "Test me on this topic and tell me what I got wrong" is active recall on demand.
- Feedback on structure. Asking how an essay plan could be better organised, or where an argument is thin, can teach — provided the student wrote it first.
Notice the common thread: in each of these, the student is still doing the thinking. AI is the tutor in the corner, not the ghostwriter.
The genuinely bad uses
The same tool, pointed the other way, actively harms learning:
Common mistake
Having it write the answer. Asking AI to write the essay, solve the problem set, or produce the assignment removes the struggle that is the learning. The work looks done; nothing was learned. Worse, it builds a dependency that collapses in the exam hall, where there's no AI — and a student who has never built the skill themselves has nowhere to fall back to.
The subtler harm is that AI makes it easy to feel productive while thinking very little. Reading a fluent AI answer produces the same false familiarity as re-reading notes — it looks like understanding, but the student couldn't reproduce it unaided. Learning lives in the effort AI is so good at removing.
The accuracy problem
A practical caution independent of how it's used: AI can be confidently wrong. It generates fluent, plausible text that is sometimes simply incorrect — a wrong date, a muddled definition, a subtly flawed worked solution — delivered with exactly the same confidence as a right answer. For a student who can't yet tell the difference, that's a real risk. Anything factual an AI produces should be checked against the textbook, the syllabus, or a teacher. Treat it as a knowledgeable but occasionally unreliable study partner, never as an authority.
Academic integrity
Finally, the rules. Schools and exam boards are actively setting policies on AI use, and they differ — some uses are fine, others count as academic misconduct with serious consequences, particularly for coursework and internal assessments that must be a student's own work. The responsibility is on the student to know and follow their school's specific rules. When unsure, the safe move is to ask the teacher before using AI on a piece of assessed work, not after.
One simple rule
All of this reduces to a single question a student can ask themselves every time: "Is this helping me think, or thinking for me?" Explaining, quizzing, generating practice, giving feedback on work you did — helping you think. Writing your answers, solving your problems, producing your assessed work — thinking for you. Stay on the right side of that line and AI is a genuinely useful addition to the study toolkit (the rest of which is here). Cross it, and it becomes the most sophisticated way yet invented to not learn.