The Simplest Test of Understanding
“If you can’t explain something simply, you don’t understand it well enough.”
Richard Feynman — Nobel Prize-winning physicist, bongo player, and one of the greatest teachers of the 20th century — had a learning method so effective that it has survived 80 years of educational research. In 2026, a Bilibili creator (月小小吖) published a 12-minute video on “Feynman Learning Method 2.0” that upgrades the original technique with AI tools and prompt engineering.
The core idea remains unchanged: teaching is the best way to learn. But the tools have evolved.
The Original 4-Step Technique (1940s)
- Choose a concept — Pick something you want to understand
- Explain it to a 12-year-old — Use simple language, no jargon
- Review and refine — Identify gaps, return to sources
- Test and simplify — Final explanation without notes
This works because of what cognitive scientists call the Protege Effect: when you prepare to teach someone else, you organize knowledge more deeply and identify gaps you would have missed through passive study.
What 2.0 Adds
| Original (1940s) | 2.0 Upgrade (2026) |
|---|---|
| Study alone | Use AI as teaching partner |
| Write on paper | Prompt engineering for deeper learning |
| One modality | Multi-modal: visual, auditory, kinesthetic |
| Manual review | AI-scheduled spaced repetition |
| Find a study buddy | AI is always available, infinitely patient |
The Science Behind It
| Principle | Research | Finding |
|---|---|---|
| Retrieval Practice | Koh et al. (2018) | Teaching enhances learning through active recall |
| Cognitive Benefits | Bargh & Schul (1980) | Organizing knowledge for instruction deepens understanding |
| Protege Effect | 2018 Study | Adopting learner-centered perspective enhances own understanding |
| Elaborative Interrogation | Cognitive Science | Asking “why” and “how” forces deeper processing |
| Generation Effect | Memory Research | Information generated from own mind is better remembered |
Learning retention rates:
- Passive learning: 5-10%
- Active learning: 50-75%
- Teaching others: 90% (Learning Pyramid model)
Three AI Prompts That Transform Learning
Prompt 1: Basic Learning Partner
Ask your AI assistant to act as a learning partner using the Feynman Technique. Have it ask what you already know, explain concepts in simple terms, use analogies, point out gaps in your understanding, and help you refine until your explanation is crystal clear.
Prompt 2: Deep Learning
Go deeper — ask the AI to create a knowledge map, explain each concept in the simplest possible language, generate analogies connected to everyday life, create quiz questions, identify common misconceptions, and suggest real-world applications.
Prompt 3: 7-Day Mastery Plan
For complete topic mastery, ask for a structured 7-day plan: foundation concepts, core principles, advanced applications, common mistakes, real-world case studies, teaching practice, and final assessment. Each day includes key concepts, simple explanations, analogies, practice exercises, and self-assessment questions.
How to Use in Daily Life
For Students:
- Study groups: Take turns teaching concepts
- Notes: Write explanations in simple language
- AI Tutor: Use Claude or ChatGPT to explain complex topics
- Record yourself teaching, then review for gaps
For Professionals:
- Meetings: Explain ideas simply to colleagues
- Documentation: Write clear, simple docs
- Onboarding: Teaching new team members deepens your own understanding
- Presentations: Use analogies for complex ideas
Daily Practice Routine:
- Morning: Pick one concept to master
- Lunch: Teach it to a colleague or AI
- Evening: Identify gaps and refine
- Weekly: Re-explain previous concepts
Key Takeaways
- Simplification is key — If you can’t explain it simply, you don’t understand it well enough
- Teaching enhances learning — The act of teaching deepens your own understanding
- AI amplifies the process — Use AI as a patient teaching partner available 24/7
- Daily practice works — Consistent 15-minute sessions beat marathon study sessions
- Identify gaps — The technique reveals what you don’t know
- Use analogies — Connect new concepts to everyday experiences
- Review regularly — Spaced repetition consolidates learning
- Apply immediately — Use new knowledge in real situations
What concept will you teach yourself today?
費曼學習法 2.0：用 AI 更快學會任何內容
最簡單的理解測試
「如果你不能簡單地解釋某件事，說明你還沒有足夠理解它。」
理查德·費曼——諾貝爾獎物理學家、邦哥鼓手、20世紀最偉大的教師之一——擁有一個如此有效的學習方法，以至於它經受了80年教育研究的考驗。2026年，一位B站創作者（月小小吖）發布了一支12分鐘影片，介紹「費曼學習法2.0」，用AI工具和提示工程升級了原始技巧。
核心理念不變：教學是學習的最佳方式。但工具已經進化。
原始4步技巧（1940年代）
- 選擇一個概念 — 挑選你想理解的內容
- 向12歲孩子解釋 — 用簡單語言，不用術語
- 回顧和改進 — 找出差距，回到資料來源
- 測試和簡化 — 不看筆記的最終解釋
這之所以有效，是因為認知科學家稱之為Protege效應的東西：當你準備教別人時，你會更深入地組織知識，並找出被動學習會忽略的差距。
2.0 增加了什麼
| 原始版（1940年代） | 2.0 升級版（2026年） |
|---|---|
| 獨自學習 | 用AI作為學習夥伴 |
| 紙上書寫 | 提示工程實現深度學習 |
| 單一模式 | 多模態：視覺、聽覺、動覺 |
| 手動複習 | AI調度的間隔重複 |
| 找學習夥伴 | AI隨時可用，無限耐心 |
背後的科學
| 原則 | 研究 | 發現 |
|---|---|---|
| 提取練習 | Koh 等人（2018年） | 教學通過主動回憶增強學習 |
| 認知益處 | Bargh 和 Schul（1980年） | 為教學組織知識能深化理解 |
| Protege 效應 | 2018年研究 | 採用以學習者為中心的觀點能增強自己的理解 |
| 精細化詢問 | 認知科學 | 問「為什麼」和「怎麼樣」迫使深度處理 |
| 生成效應 | 記憶研究 | 從自己頭腦中生成的資訊更容易記住 |
學習保留率：
- 被動學習：5-10%
- 主動學習：50-75%
- 教導他人：90%（學習金字塔模型）
改變學習的三個 AI 提示
提示 1：基礎學習夥伴
要求你的AI助手使用費曼技巧作為學習夥伴。讓它問你已經知道什麼，用簡單術語解釋概念，使用類比，指出你理解中的差距，並幫你改進直到解釋完全清晰。
提示 2：深度學習
更深入——要求AI創建知識地圖，用最簡單的語言解釋每個概念，生成連結到日常生活的類比，創建測驗問題，識別常見誤解，並建議現實世界的應用。
提示 3：7天精通計劃
要完全掌握某個主題，要求制定結構化的7天計劃：基礎概念、核心原則、進階應用、常見錯誤、現實世界案例研究、教學練習和最終評估。每天包括關鍵概念、簡單解釋、類比、練習和自我評估問題。
如何在日常生活中使用
給學生：
- 學習小組：輪流教授概念
- 筆記：用簡單語言寫解釋
- AI家教：使用 Claude 或 ChatGPT 解釋複雜主題
- 錄下自己教學，然後回顧差距
給專業人士：
- 會議：向同事簡單解釋想法
- 文件：撰寫清晰、簡單的文件
- 入職培訓：教導新團隊成員能深化自己的理解
- 演示：使用類比解釋複雜想法
每日練習流程：
- 晨間：選擇一個要掌握的概念
- 午餐：教給同事或AI
- 晚間：找出差距並改進
- 每週：重新解釋之前的概念
關鍵要點
- 簡化是關鍵 — 如果你不能簡單地解釋它，說明你還沒有足夠理解它
- 教學增強學習 — 教學行為能深化自己的理解
- AI 放大過程 — 使用AI作為24/7可用的耐心教學夥伴
- 每日練習有效 — 持續15分鐘的課程勝過馬拉松式的學習
- 找出差距 — 這個技巧能揭示你不知道的內容
- 使用類比 — 將新概念連結到日常經驗
- 定期複習 — 間隔重複鞏固學習
- 立即應用 — 在現實情況中使用新知識
你今天將教自己什麼概念？
References:
- Richard Feynman, “Surely You’re Joking, Mr. Feynman!” (1985)
- Koh, A. W., et al. “The Protege Effect in Learning” (2018)
- Bargh, J. & Schul, E. “On the Cognitive Mediation of Learning” (1980)
- Learning Pyramid Model — National Training Laboratories
- Bilibili Video: 月小小吖 — “費曼學習法2.0” (2026)