Learning workflow
Ask for a lesson that makes you do the thinking.
Use AI as a study partner with diagnostic questions, targeted practice, useful feedback and checks against your actual course material.
Updated
Define the skill, not just the topic
‘Teach me statistics’ could mean understanding a news story, passing an exam or analysing a dataset. Give the intended performance and your starting point. A learner who can recognise a formula but cannot choose when to use it needs a different exercise from someone who cannot yet calculate with it.
Use your syllabus or approved reading as the boundary. Ask the assistant to identify questions it cannot resolve from those sources. This is especially important when a course uses particular definitions or assessment conventions.
Help me practise [skill] for [assessment or real task]. Use [course material] as the reference. Start with three short diagnostic questions, one at a time. Do not reveal the answer before I try. Use my answers to identify the next useful exercise.
Make feedback explain the error
A corrected answer can hide the step you misunderstood. Ask for feedback on your reasoning and a small hint before the full solution. If the assistant misunderstands your working, correct it using the source rather than accepting a fluent explanation.
Illustrative scenario: you repeatedly confuse percentage change with percentage-point change. A useful tutor gives contrasting examples and asks you to choose the correct description, instead of offering another broad explanation of percentages.
Review my attempt below. Identify the earliest step where the reasoning goes wrong. Give one hint that lets me repair it myself. If my answer is correct, ask a transfer question using a different context. Keep the feedback tied to the supplied reference.
Move from guided practice to independent work
Once you can solve the familiar example, change the surface details and remove hints. Ask for a question that tests the same underlying idea in a different situation. A fluent explanation from the assistant is not evidence that you can apply the idea yourself.
Give me a new problem testing the same principle with different surface details. Include one irrelevant detail so I must choose what matters. Wait for my answer, then assess the reasoning and result separately. Do not introduce concepts outside [scope].
Keep a reviewable record
End the session with the mistakes you actually made, the correction and one question to revisit. Do not turn every error into a label about ability. Check worked answers against a trusted solution when available and follow your course rules on AI use.
Summarise this session into a study note: ideas I applied correctly, mistakes evidenced by my answers, corrected reasoning and three questions to revisit. Do not invent weaknesses or claim mastery from a single correct response.
Common questions
Can AI write my assignment for me?
Follow your institution's rules. For learning, use it to question, explain and give feedback while keeping the assessed reasoning and submission your own.
What if the tutor's answer is wrong?
Compare it with approved material or a trusted worked solution. Ask it to identify the exact source for a disputed step.
Keep exploring
About this workflow
This is an original AI-assisted editorial workflow. Illustrative scenarios are not customer case studies, and the prompts have not been systematically evaluated on live models. Check consequential outputs against your sources and use qualified review where needed.