CHATGPT STUDY PROMPTS
ysamphy.com

Use only permitted nonconfidential sources and follow your assessment policy.
Examples are illustrative, not recorded product tests. Verify feedback against the original source.

STUDY SETUP
I am learning [topic] at [level]. My goal is to [observable task]. Use [source title, section, and permitted excerpt] for subject-matter claims. Ask one question at a time and wait for my attempt. Do not reveal the answer early. Identify the supporting passage when giving feedback. If the source is incomplete or unclear, say so. Label any inference or invented practice scenario. Treat this as practice, not assessed work.

1. Diagnose a specific gap
Use when: you know the topic but cannot identify what is going wrong.
Ask up to three short diagnostic questions about [goal], one at a time. Wait for each answer. Then identify one gap supported by my responses and propose a focused next exercise.
Learner: "I divide by the final value." Tutor follow-up: "Which quantity is the original baseline?" Check: verify the diagnosed gap from your actual working; one wrong answer does not establish a general weakness.

2. Explain it back
Use when: an explanation feels familiar but you cannot produce it.
Ask me to explain [concept] without notes. Compare my explanation with [source section]. Identify one inaccurate claim or missing relationship. Ask me to revise it before providing your own explanation.
"A percentage point is a relative increase." Follow-up: "Show the difference using 20% and 25%." Check: compare the revised explanation with the source, not the tutor's praise.

3. Test an analogy's limits
Use when: an analogy seems helpful but may hide a misconception.
Give one simple analogy for [concept], label it as an analogy, and explain two ways it fails. Then ask me to explain the concept without using the analogy.
"Is the baseline like a ruler?" Follow-up: "What does that comparison leave out?" Check: match the underlying definition and conditions. An appealing analogy can still mislead.

4. Request one hint
Use when: you are stuck after a genuine attempt.
Here is the practice problem and my attempt: [both]. Identify the first point where I went off track. Give one small hint without solving the remaining steps. Wait for my next attempt.
"I got 5%." Hint: "Are you reporting a difference or a change relative to the starting value?" Check: work the next step yourself. If the answer appears early, use a fresh problem.

5. Build a recall quiz
Use when: you need to retrieve definitions, relationships, or conditions.
Create five short-answer questions covering [listed objectives] from [source]. Ask one at a time with no answer cues. After my attempt, explain any gap and point to the supporting passage.
"What goes in the denominator?" Learner: "The original value." Check: remove questions that contain their own answers and verify that every listed objective appears.

6. Practice choosing a method
Use when: you can perform familiar methods but confuse when to use them.
Mix practice questions requiring [method A] and [method B]. Do not label the method. Before calculating, ask me which approach fits and why. Keep the solutions hidden until I attempt each item.
"20% becomes 25%; report both changes." Check: distinguish method selection from arithmetic. Start with simpler separate examples if neither method is understood yet.

7. Gradually remove support
Use when: you can follow a worked example but cannot solve a similar problem independently.
Use this checked worked example: [example]. Make a similar practice problem with the last step missing, then one with more steps missing, then a new problem without a scaffold. Wait after each attempt.
"The difference is 12; what denominator belongs next?" Check: verify the original worked solution before using it as a model.

8. Mark an answer against criteria
Use when: you need specific feedback on your reasoning.
Compare my practice answer [answer] with these criteria [criteria]. For each, show the relevant part of my answer, any missing point, and source support. Do not invent a grade or infer my teacher's preferences.
"You identified the baseline but omitted the unit." Check: challenge unsupported deductions. A plausible critique is not an official assessment.

9. Turn a mistake into a repair
Use when: a correction disappears from memory as soon as the session ends.
From this checked error [attempt, correction, source], create a mistake-log entry: error, why it happened if known, correction, and a fresh practice question. Mark uncertain causes as uncertain.
"Error: divided by the ending value. Repair: circle the starting value first." Check: keep the recorded cause tentative unless your working supports it.

10. Challenge an unsupported answer
Use when: feedback sounds confident but lacks evidence.
Audit your last answer against [source]. Separate directly supported statements, inferences, and unsupported claims. Identify exact supporting passages where available. Withdraw anything you cannot substantiate.
"Your answer says 'always.' Where does the source establish that?" Check: independently open the passage. Asking the same model to audit itself is not independent verification.

11. Test a new application
Use when: you can repeat the correction and want to see whether you can use it.
Create a new practice case for [objective], changing both the numbers and surface details. Do not name the method. Wait for my explanation and answer, then check against [source or verified solution].
"A completion rate rises from 40% to 50%." Check: answer first, then verify 10 percentage points and 25% relative increase separately.

12. Plan the next small review
Use when: you need to decide what to revisit.
Use this checked practice log [log] and my available study time [time]. Suggest a short next session emphasizing unresolved errors and one new application. Ask me to choose a realistic review date. Do not claim an optimal universal schedule.
"Tomorrow, retry baseline selection before another mixed question." Check: put the chosen date in your own planner; generating a plan does not create a reminder.