Make Claude Quiz You
Swipe to show menu
Reading is not studying
Re-reading an explanation produces a strong feeling of competence and very little retention. Retrieving the information from your own head — with the page closed — produces the opposite: it feels harder and works far better. This is the most robust result in the study of learning.
Active recall (retrieval practice)
Deliberately pulling information out of memory without looking at the source. The act of retrieval is what strengthens the memory — the effort is doing the work, not the exposure. Recognising an answer when you see it is a different and much weaker operation than producing it.
Flip the roles
The default relationship is: you ask, Claude explains. Invert it.
1. The Feynman flip
The Feynman technique
Explain a concept in plain language as though to a beginner. Wherever you stall, hedge, or reach for jargon, you've located a gap in your own understanding. The gaps are the output — the explanation itself is just the instrument that finds them.
I'm going to explain [CONCEPT] to you as if you're a beginner.
Grade me strictly out of 10 on:
- accuracy
- completeness
- whether a beginner could actually follow it
Then list, specifically, what I left out or stated too confidently.
Do not be encouraging. I want the gaps.
My explanation:
[write it yourself, no looking]
Do not be encouraging is load-bearing. Without it you get praise, and praise is not
information.
2. Question generation
Generate 8 questions on [TOPIC] for a learner at [LEVEL]:
- 3 recall (definitions, syntax)
- 3 application (here's a scenario, what happens)
- 2 diagnostic (here's broken code/reasoning, what's wrong)
Ask them ONE at a time. Wait for my answer. Then tell me if I'm right,
what I missed, and only then move on.
The one-at-a-time constraint matters. A list of eight questions becomes something you skim. One question with a wait becomes a test.
3. Flashcards
Turn my notes below into 12 flashcards, front/back, as a two-column table.
Rules: one idea per card, no card answerable with yes/no, and front side
must be a question I could be asked in an interview.
Notes:
[paste]
Spacing Beats Cramming
Spaced repetition
Distributing review across increasing intervals rather than massing it into one session. The same total study time produces substantially better long-term retention, because each retrieval after partial forgetting strengthens the memory more than an easy one does.
Same total minutes, far better retention, just redistributed:
| When | What you do | Minutes |
|---|---|---|
| Day 0 | Watch, take notes, make cards | — |
| Day 1 | Recall from cards. Mark every miss. | 10 |
| Day 3 | Only the misses. Then Feynman-explain the topic. | 15 |
| Day 7 | Full pass. Ask Claude for 3 new application questions. | 10 |
The Day 7 "new questions" step is important — reusing the same cards tests your memory of the cards, not of the concept.
The honesty setting
Claude defaults to being supportive, which is pleasant and useless for assessment. Override it explicitly:
Grade me harshly. Assume I want to be corrected, not encouraged.
If my answer is 70% right, tell me about the 30%.
Thanks for your feedback!
Ask AI
Ask AI
Ask anything or try one of the suggested questions to begin our chat
Make Claude Quiz You
Reading is not studying
Re-reading an explanation produces a strong feeling of competence and very little retention. Retrieving the information from your own head — with the page closed — produces the opposite: it feels harder and works far better. This is the most robust result in the study of learning.
Active recall (retrieval practice)
Deliberately pulling information out of memory without looking at the source. The act of retrieval is what strengthens the memory — the effort is doing the work, not the exposure. Recognising an answer when you see it is a different and much weaker operation than producing it.
Flip the roles
The default relationship is: you ask, Claude explains. Invert it.
1. The Feynman flip
The Feynman technique
Explain a concept in plain language as though to a beginner. Wherever you stall, hedge, or reach for jargon, you've located a gap in your own understanding. The gaps are the output — the explanation itself is just the instrument that finds them.
I'm going to explain [CONCEPT] to you as if you're a beginner.
Grade me strictly out of 10 on:
- accuracy
- completeness
- whether a beginner could actually follow it
Then list, specifically, what I left out or stated too confidently.
Do not be encouraging. I want the gaps.
My explanation:
[write it yourself, no looking]
Do not be encouraging is load-bearing. Without it you get praise, and praise is not
information.
2. Question generation
Generate 8 questions on [TOPIC] for a learner at [LEVEL]:
- 3 recall (definitions, syntax)
- 3 application (here's a scenario, what happens)
- 2 diagnostic (here's broken code/reasoning, what's wrong)
Ask them ONE at a time. Wait for my answer. Then tell me if I'm right,
what I missed, and only then move on.
The one-at-a-time constraint matters. A list of eight questions becomes something you skim. One question with a wait becomes a test.
3. Flashcards
Turn my notes below into 12 flashcards, front/back, as a two-column table.
Rules: one idea per card, no card answerable with yes/no, and front side
must be a question I could be asked in an interview.
Notes:
[paste]
Spacing Beats Cramming
Spaced repetition
Distributing review across increasing intervals rather than massing it into one session. The same total study time produces substantially better long-term retention, because each retrieval after partial forgetting strengthens the memory more than an easy one does.
Same total minutes, far better retention, just redistributed:
| When | What you do | Minutes |
|---|---|---|
| Day 0 | Watch, take notes, make cards | — |
| Day 1 | Recall from cards. Mark every miss. | 10 |
| Day 3 | Only the misses. Then Feynman-explain the topic. | 15 |
| Day 7 | Full pass. Ask Claude for 3 new application questions. | 10 |
The Day 7 "new questions" step is important — reusing the same cards tests your memory of the cards, not of the concept.
The honesty setting
Claude defaults to being supportive, which is pleasant and useless for assessment. Override it explicitly:
Grade me harshly. Assume I want to be corrected, not encouraged.
If my answer is 70% right, tell me about the 30%.
Thanks for your feedback!
