Discussion drills

1. Diagnose 1

   A student uses Claude to produce a set of revision notes summarising a history unit. The notes are accurate and well-structured. The student reads them three times. On the exam, they cannot recall specific dates or causal chains that were in the notes.

   Identify the diligence failure. It is not a failure of note quality. Name the specific cognitive process that was skipped and why reading the notes three times did not substitute for it.

   ▼ Teacher hint
2. Predict 2

   You ask Claude: "Quiz me on the French Revolution." Claude produces 10 questions, you answer them, Claude marks them correct or incorrect. Predict: what does this method reliably build in the student? What does it not build? What would you change in the design to make it more effective?

   ▼ Teacher hint
3. Construct 3

   Design a diligent AI revision session for the topic "causes of WWI." Three hard requirements: (a) the student must produce, not read; (b) the session must reveal gaps the student did not know they had; (c) the student must close gaps themselves, not by reading Claude's explanation. Write the full sequence of prompts — not the plan, the actual prompts.

   ▼ Teacher hint
4. Judge 4

   "Using AI to write your first essay draft and then heavily editing it yourself is a legitimate form of creation diligence." Evaluate this claim. What does "heavily editing" need to mean for the claim to be true? Where is the hard line?

   ▼ Teacher hint
5. Compare 5

   AI as note-taker vs AI as examiner — what does each model build in the student, and what does each fail to build? Which produces better long-term recall, and what is the mechanism that explains the difference?

   ▼ Teacher hint
6. Falsify 6

   "If you always verify AI output, you are being diligent." Make the strongest case that verification alone is not diligence. Include a scenario where verification is complete — every fact is checked, every claim is accurate — and learning is zero.

   ▼ Teacher hint