Student Study Pack

Using AI to learn faster, without ending up unable to do the thing you were supposed to learn.

The line

Using AI to understand something is learning. Using it to produce work you submit as your own is not.

Check your institution's policy — they vary widely and the penalties are serious. Where the policy is unclear, ask before you submit, not after.

The deeper problem is not getting caught. It is that AI can produce work you cannot defend in a viva, an exam, or a job interview, and you will not know the gap exists until it matters.

The honest test

Before submitting anything:

Could I explain every sentence of this to my lecturer, without notes, and answer a follow-up question?

If no, do not submit it. That is the same standard an exam applies.

Understanding

Explain at my level

Explain [CONCEPT] to me.
I already understand: [WHAT YOU KNOW]
I do not have background in: [WHAT YOU LACK]

Use one analogy, then drop it and be precise.
End with the thing most students get wrong about this.

The prerequisite check

I want to understand [TOPIC] but I keep getting lost.
What do I need to understand first?
Ask me two questions to check whether I already do. Wait for my answers
before explaining anything.

This one saves enormous amounts of time. Most confusion is a missing prerequisite, not the topic itself.

Interrogating a paper

I have uploaded [PAPER]. Do not summarise it.

1. What is the central claim?
2. What evidence supports it?
3. What is the methodology, in plain English?
4. What could go wrong with that methodology?
5. What would a critic say?
6. Quote the single most important sentence.

Point 4 is where the marks are. Most students describe a study; the ones who evaluate it get the higher grades.

Testing yourself

This is where AI genuinely outperforms other study methods.

The quiz

Ask me 10 questions on [TOPIC], one at a time, starting easy.
After each answer, tell me what I got right, what I missed,
and then ask the next.
Do not give me the next question until I answer.
Do not tell me the answer before I try.

The Feynman test

I am going to explain [TOPIC] as if teaching it.
Listen, then tell me: what did I get wrong, what did I skip, and where
did I use a term I clearly do not fully understand?

My explanation: [YOUR EXPLANATION, WITHOUT LOOKING ANYTHING UP]

The most effective study technique on this page. Explaining exposes gaps that re-reading hides.

Spaced retrieval

Create a five-question retrieval set mixing:
- 2 from [THIS WEEK'S TOPIC]
- 2 from [LAST MONTH]
- 1 from [LAST TERM]

Answers separately so I can test myself first.

Flashcards

Create flashcards for [TOPIC] from this material: [PASTE]

Format: Question | Answer
Rules:
- One idea per card
- Questions that require recall, not recognition
- No "what is X?" where the answer is a definition I could guess
- Include 3 cards on common misconceptions, phrased as questions
- 15 cards maximum — more than that and I will not review them

Writing, legitimately

Planning

My essay question: [QUESTION]
My argument: [YOUR ARGUMENT, IN YOUR WORDS]
Evidence I have: [YOUR SOURCES]

Critique my argument. Where is it weak? What counter-argument
am I not addressing? What evidence would strengthen it?

Do not write any of it for me.

Feedback on your own draft

Here is my draft: [PASTE]
The question: [QUESTION]
Marking criteria: [PASTE IF YOU HAVE IT]

Tell me:
- Does it answer the question asked?
- Where is the argument unclear?
- Where do I assert without evidence?
- What would a marker query?

Do not rewrite anything. Tell me what to fix.

The "do not rewrite" instruction is what keeps this legitimate. Feedback on your work is what a tutor does. Having it rewritten is not.

Sources — the trap

AI invents citations. Realistic authors, realistic journals, realistic years, realistic DOIs. Students have submitted these and failed.

• Never cite anything you have not opened.
• Use AI to find search terms, not sources.
• If it names a paper, search for it yourself. Ten seconds and it does not exist? It does not exist.
• Read what you cite. A summary of an abstract is not a reading of the paper, and a marker can tell.

What this costs you if you overuse it

The real risk is not detection.

• You cannot write under exam conditions, because you have not practised.
• You cannot defend your work in a viva or an interview.
• You have not built the thinking, and later modules assume you did.
• You cannot judge whether the output is good, because you never developed the judgement.

The students who get the most from AI use it to check their understanding, not to replace it.

Your rules

My institution's policy says
I use AI for
I never use AI for
How I disclose it, if required