The Best ‘Teach, Don’t Summarize’ Prompt

Why Teaching Beats Summaries

Here’s a scenario you know well: you read an article, ask AI to summarize it, skim the summary, and move on. An hour later, you can’t remember a thing. The summary gave you the gist, but the gist evaporated.

The problem isn’t AI. It’s the prompt. “Summarize this” tells AI to compress information. But compression doesn’t equal comprehension. You can read a perfect summary without understanding the underlying ideas at all.

The teach don’t summarize prompt flips this dynamic. Instead of asking AI to explain the text to you, you explain it to AI — and AI tells you what you got right, wrong, and missed. This is the difference between passive consumption and active processing.

Learning science calls this the generation effect: producing information from memory strengthens retention more than passively receiving it. When you try to explain something, you discover what you actually understand versus what you only think you understand.

The Base Prompt: Explain It Back

The core prompt above (PR033) is deceptively simple. You write your explanation first — without looking at the text — then paste it into the prompt. AI evaluates your attempt against what the text actually said.

This exposes the gap between recognition and recall. Recognition means seeing something and thinking “yes, I knew that.” Recall means producing it from memory without cues. Most people confuse the two and overestimate their understanding.

The prompt forces recall. You can’t fake it. If you can only repeat phrases from the text but can’t rephrase concepts in your own words, the AI will notice. If you missed the author’s main point entirely, that’ll surface too.

7 Variations for Different Learning Styles

The base prompt works for general comprehension checks. But different situations call for different approaches. Here are seven variations you can adapt:

1. The Feynman Variation

“I’m going to explain this concept as if teaching a 12-year-old. Tell me where my explanation would confuse them or where I’m using jargon I haven’t defined.”

2. The Connection Variation

“I’m going to explain how this relates to [another concept I know]. Tell me if my analogy is accurate or if it breaks down at certain points.”

3. The Application Variation

“I’m going to explain how I would apply this concept to [specific situation]. Tell me if I’m applying it correctly or misunderstanding how it works.”

4. The Debate Variation

“I’m going to argue against the author’s main point. Tell me if I’m addressing their actual argument or attacking a strawman.”

5. The Prediction Variation

“Based on what I read, here’s what I predict about [related topic]. Tell me if my prediction follows logically from the text or if I’m overreaching.”

6. The Summary + Gaps Variation

“Here’s my summary. But more importantly, tell me what the author would say I’m missing — not just facts, but nuances and qualifications.”

7. The Teach-Back Chain

“I explained this to you. Now quiz me with 3 questions to test whether my understanding is solid or superficial. Make the questions progressively harder.”

Each variation targets a different failure mode. The Feynman variation catches jargon-hiding. The Connection variation prevents false analogies. The Debate variation ensures you understood the actual argument before criticizing it.

When to Use Which Variation

Match the variation to your reading goal:

For conceptual understanding, use the base prompt or the Feynman variation. These check whether you grasp the core ideas well enough to explain them simply.

For critical reading, use the Debate variation. This ensures you’re engaging with the author’s actual claims rather than a version you invented. Pair it with the Socratic Reading Prompts for deeper analysis.

For practical application, use the Application or Prediction variations. These test whether you can transfer knowledge to new contexts — the gold standard for real learning.

For exam preparation, use the Teach-Back Chain. This simulates test conditions and reveals gaps before the actual exam. Combine it with prompts from our article understanding guide for comprehensive prep.

Common Mistakes to Avoid

Mistake 1: Looking at the text while explaining. This defeats the purpose. You’re testing recall, not recognition. Close the article before writing your explanation.

Mistake 2: Explaining too briefly. A one-sentence explanation won’t reveal much about your understanding. Aim for a paragraph at minimum — enough detail for AI to assess nuance.

Mistake 3: Ignoring the feedback. If AI says you missed something, don’t just acknowledge it and move on. Go back to the text, find what you missed, and try explaining again.

Mistake 4: Only using this for difficult texts. The teach-back method works for everything — including content you think you already understand. Often, “easy” articles reveal surprising gaps when you try to explain them.

The AI Reading Prompts Library contains more tools for different reading challenges. But this one — the simple act of explaining before asking — might be the highest-leverage change you can make to how you read with AI.

Ready to try it? Pick something you read recently. Write your explanation. Paste it into the prompt. See what you actually learned — and what you only thought you did.