用乔哈里窗(Johari Window)

解锁你的AI提问力

🔒

Using the Johari Window

Unlock Your AI Prompting Power

🔒

思维转换的艺术 The Art of Mindset Transformation

"Talk is cheap,
show me the code"

—— by Linus Torvalds

"Code is cheap,
show me the talk"

一个问题：你与AI的对话，停留在哪个层次？ A Question: At What Level is Your Dialogue with AI?

📝 任务 📝 Task
写诗，关于夏天 Write a poem about summer
💬 普通提问 💬 Basic Prompting
"请帮我写一首关于夏天的诗" "Please write me a poem about summer"
✅ 高级用法 ✅ Advanced Usage
"我需要一首现代诗，表达对夏天的歌颂，风格幽默诙谐，不要故作惆怅，14行以内" "I need a modern poem celebrating summer, with a humorous and witty style, no forced melancholy, within 14 lines"

我们与AI的四种认知关系 Four Cognitive Relationships Between Us and AI

AI知道 AI Knows
AI不知道 AI Doesn't Know
我知道 I Know
我不知道 I Don't Know
(0,0)
I

象限I 公开区 Quadrant I: Open Area

简单说 Just Ask

例："今天天气怎么样？"
"帮我把这篇文章精简一下" e.g. "What's the weather today?"
"Help me condense this article"

(+X, +Y)

II

象限II 隐藏区 Quadrant II: Hidden Area

喂模式 Feed Mode

例："我公司的销售数据是..."
"帮我看看这份体检报告的关键注意事项" e.g. "My company's sales data is..."
"Help analyze key points in this medical report"

(+X, -Y)

III

象限III 盲点区 Quadrant III: Blind Spot

提好问题 Ask Smart Questions

例："Macbook M4和M4 pro的性能差异"
"如何提高团队协作效率？" e.g. "Performance differences between MacBook M4 and M4 Pro"
"How to improve team collaboration?"

(-X, +Y)

IV

象限IV 未知区 Quadrant IV: Unknown Area

开放聊 Open Exploration

例："我们一起探讨一下
未来AI的发展可能性" e.g. "Let's explore together
the future possibilities of AI"

(-X, -Y)

公开象限 (AI知道，你也知道) Open Area (AI Knows, You Know)

核心关键词：自动化 (Automation) & 效率 (Efficiency) Keywords: Automation & Efficiency

AI角色： AI Role: 完美的高级助理 Perfect Senior Assistant

应用场景： Application Scenarios:

  • 代码重构、格式转换 Code refactoring, format conversion
  • 信息查询、语言翻译 Information queries, language translation
  • 编写单元测试 Writing unit tests

关键行动： Key Actions: 授权 (Delegate) & 执行 (Execute) Delegate & Execute

演讲者备注： Speaker Notes: 这是我们最熟悉的区域。所有你明确知道怎么做，AI也知道怎么做的事情，都应该毫不犹豫地交给它，把你的时间解放出来。 This is the area we're most familiar with. All tasks that you clearly know how to do and AI also knows how to do should be handed over to it without hesitation, freeing up your time.

隐藏象限 (你知道，AI不知道) Hidden Area (You Know, AI Doesn't Know)

核心关键词：个性化 (Personalization) & 知识注入 (Injection) Keywords: Personalization & Knowledge Injection

AI角色： AI Role: 需要你来"喂养"的学生 Student that needs your "feeding"

应用场景： Application Scenarios:

  • 上传个人病历，进行健康咨询 Upload personal medical records for health consultation
  • 输入公司设计规范，生成UI代码 Input company design specifications to generate UI code
  • 投喂你的项目文档，让AI帮你写总结 Feed your project documents to let AI help write summaries

关键行动： Key Actions: 投喂 (Feed) & 训练 (Train) Feed & Train

演讲者备注： Speaker Notes: 这是你创造独特价值的地方。AI的通用知识库里没有你的个人经验、项目数据和团队规范。通过'投喂'，你正在把一个通用AI，训练成你的专属AI。 This is where you create unique value. AI's general knowledge base doesn't contain your personal experience, project data, and team standards. Through 'feeding', you're training a general AI into your exclusive AI.

盲点象限 (AI知道，你不知道) Blind Spot (AI Knows, You Don't Know)

核心关键词：学习 (Learning) & 认知拓展 (Extension) Keywords: Learning & Cognitive Extension

AI角色： AI Role: 你的"外置大脑" Your "External Brain"

应用场景： Application Scenarios:

  • "解释一下什么是'边际效用递减'？" "Explain what 'diminishing marginal utility' means?"
  • "我只用过Vue，React Hooks的核心思想是什么？" "I've only used Vue, what's the core concept of React Hooks?"
  • "部署Web应用时，Vercel和Cloudflare Pages的优劣对比？" "What are the pros and cons of Vercel vs Cloudflare Pages for web app deployment?"

关键行动： Key Actions: 提问 (Ask) & 探索 (Explore) Ask & Explore

演讲者备注： Speaker Notes: 这个象限能帮你快速突破知识盲区。把AI当作一个24小时在线、不知疲倦的专家导师，保持好奇心，大胆地问！ This quadrant helps you quickly break through knowledge blind spots. Treat AI as a 24/7 online, tireless expert mentor, maintain curiosity, and ask boldly!

未知象限 (你不知道，AI也不知道) Unknown Area (You Don't Know, AI Doesn't Know)

核心关键词：创新 (Innovation) & 涌现 (Emergence) Keywords: Innovation & Emergence

AI角色： AI Role: 平等的共创伙伴 (Co-pilot) Equal Co-creation Partner (Co-pilot)

应用场景： Application Scenarios:

  • 头脑风暴新产品、新功能 Brainstorm new products and features
  • 探索新的艺术风格、小说大纲 Explore new artistic styles, novel outlines
  • 挑战现有科学难题（如AlphaDev发现新算法） Challenge existing scientific problems (like AlphaDev discovering new algorithms)

关键行动： Key Actions: 共创 (Co-create) & 迭代 (Iterate) Co-create & Iterate

演讲者备注： Speaker Notes: 这是最激动人心的区域，是真正'1+1>2'的地方。在这里，没有标准答案，你和AI通过高质量的对话，共同探索可能性的边界。 This is the most exciting area, where true '1+1>2' happens. Here, there are no standard answers, you and AI explore the boundaries of possibility through high-quality dialogue.

进阶心法 Advanced Strategies

一个核心目标： One Core Goal:

不断压缩"盲点"和"隐藏"象限，将"未知"的灵感，转化为"公开"的成果。 Continuously compress "Blind Spot" and "Hidden" quadrants, transforming "Unknown" inspiration into "Open" achievements.

❓
提问的工程学 The Engineering of Questioning
C.R.A.F.T. 框架 C.R.A.F.T. Framework
🧠
你的第二大脑 Your Second Brain
知识库构建 Knowledge Base Building
🤖
AI复仇者联盟 AI Avengers Alliance
多模型混用 Multi-Model Integration

思维转变：从"信息索取者"到"对话架构师" Mindset Shift: From "Information Seeker" to "Dialogue Architect"

搜索引擎时代： Search Engine Era: 我们是索取者，期待找到现成答案。 We are seekers, expecting to find ready-made answers.

⬇️

AI大模型时代： AI Large Model Era: 我们是架构师，通过搭建框架，让AI生成答案。 We are architects, building frameworks to let AI generate answers.

💡 核心观点：一个好的问题，本质上是为AI构建了一个优质的思考脚手架。 Core Insight: A good question essentially builds a high-quality thinking scaffold for AI.

演讲者备注： Speaker Notes: 在你问出问题之前，请先转变心态。你不是在搜索，你是在编程——用语言对AI进行编程。你给的指令越清晰，程序运行的结果就越好。 Before asking a question, first shift your mindset. You're not searching, you're programming—programming AI with language. The clearer your instructions, the better the program runs.

黄金提问框架：C.R.A.F.T. Golden Prompting Framework: C.R.A.F.T.

C
Context
对话的"地基"
你在哪，我是谁？ The Foundation
Where are you, who am I?
R
Role
为AI戴上"职业镜头"
你是谁？ Professional Lens for AI
Who are you?
A
Action
"做什么"和"什么不能做"？ What to do & constraints?
F
Format
给我看"样子"
别只说"样子" Show me examples
Don't just describe
T
Target
对话的"指南针"
我们要去哪？ The Compass
Where are we going?

无效提问 vs. C.R.A.F.T.提问 Ineffective vs. C.R.A.F.T. Prompting

❌ 无效提问 ❌ Ineffective Prompting
"我的React代码不工作了，帮我看看。" "My React code isn't working, help me check it."

结果：得到一个通用的、可能无效的模糊答案。 Result: Gets a generic, potentially ineffective vague answer.

✅ C.R.A.F.T.提问 ✅ C.R.A.F.T. Prompting
(C) 我在用Hooks写计数器。
(R) 你是React辅导专家。
(A) 帮我修bug，但只能用useState。
(F) 按"问题定位-修正代码-原理解释"的范例回复。
(T) 我的目标是彻底搞懂它。 (C) I'm writing a counter with Hooks.
(R) You are a React tutoring expert.
(A) Help me fix the bug, but only use useState.
(F) Reply following the "Problem identification - Code correction - Principle explanation" format.
(T) My goal is to thoroughly understand it.

结果：得到一个量身定制、解释清晰、能真正解决问题的完美答案。 Result: Gets a tailored, clearly explained, perfect answer that truly solves the problem.

演讲者备注： Speaker Notes: 大家可以直观感受一下，同样一个问题，因为提问方式的不同，得到的结果天差地别。右边的提问，几乎是在'逼迫'AI给出一个满分答案。 You can intuitively feel that for the same question, different prompting methods lead to vastly different results. The right-side prompting almost 'forces' AI to give a perfect answer.

知识库 与 AI联盟 Knowledge Base & AI Alliance

心法二：知识库是你的第二大脑 Strategy 2: Knowledge Base as Your Second Brain

为什么？ Why? 将你的"隐藏象限"转化为AI可用的燃料。 Transform your "Hidden Area" into AI-usable fuel.

怎么做？ How? 使用Notion, 2Brain等工具，沉淀你的笔记、代码、文档。 Use tools like Notion, 2Brain to accumulate your notes, code, and documents.

🧠💾

心法三：构建你的"AI复仇者联盟" Strategy 3: Build Your "AI Avengers Alliance"

为什么？ Why? 没有一个AI是万能的。 No single AI is omnipotent.

怎么做？ How? 任务拆解：用GPT-4o头脑风暴，用Claude写代码，用Midjourney做图。 Task breakdown: Use GPT-4o for brainstorming, Claude for coding, Midjourney for images.

🤖🤖🤖

Workshop: 小组任务 Workshop: Group Task

背景： Background: 为"小蜜蜂应用商店"设计并实现一个新的 Design and implement a new "AI智能评测" "AI Smart Evaluation" 功能。 feature for "Little Bee App Store".

任务： Task: 请用"乔哈里窗"四象限，规划你们将如何利用AI完成这个任务。 Please use the "Johari Window" four quadrants to plan how you will leverage AI to complete this task.

  • 公开象限： Open Area: 哪些任务可直接让AI生成？ Which tasks can be directly generated by AI?
  • 隐藏象限： Hidden Area: 你们需要"喂"给AI什么独有信息？ What unique information do you need to "feed" to AI?
  • 盲点象限： Blind Spot: 你们需要向AI请教哪些问题？ What questions do you need to ask AI?
  • 未知象限： Unknown Area: 你们希望和AI一起头脑风暴出哪些创新点？ What innovative ideas do you hope to brainstorm with AI?

⏰ 时间：8分钟讨论，2分钟分享 Time: 8 minutes discussion, 2 minutes sharing

演讲者备注： Speaker Notes: 现在，我们把理论用于实践。请大家分组讨论，这不是一个技术问题，而是一个策略问题。思考一下如何最大化地利用AI这个伙伴。 Now, let's put theory into practice. Please discuss in groups. This is not a technical problem, but a strategic one. Think about how to maximize the use of AI as a partner.

Takeaway: 你的收获 Takeaway: Your Gains

🗺️ 一个地图： One Map: 乔哈里窗，帮你识别人机协作的状态。 Johari Window helps you identify human-AI collaboration states.

🎯 两个动作： Two Actions: 高质量的"提问"和"投喂"。 High-quality "Questioning" and "Feeding".

🚀 一个目标： One Goal: 与AI共创，在"未知象限"中创造价值。 Co-create with AI, creating value in the "Unknown Area".

Q&A

"在AI时代，决定你高度的，不再是你拥有多少知识，而是你与智能体建立认知连接、并共同创造新价值的能力。" "In the AI era, what determines your height is no longer how much knowledge you possess, but your ability to establish cognitive connections with intelligent agents and co-create new value together."
1 / 15