Learning and Development in the AI Era – Navigating Your Intelligent “Co-pilot”#


The correct role of AI tools in product development: Augmentation, not Replacement	Structured approach to Prompt Engineering: the C-R-E-F framework	Human-AI quality control workflow: The Sandwich Method	Red lines for AI use: privacy, intellectual property, and professional accountability	Connection to PMBOK: Stewardship, Team Performance Domain

AI 工具在产品开发中的正确角色定位：增强(Augmentation)而非替代(Replacement)

Prompt Engineering 的结构化方法：C-R-E-F 框架

人机协作的质量控制工作流：三明治工作法（The Sandwich Method）

AI 使用的红线规则：隐私、知识产权与专业责任

关联 PMBOK：尽职管理(Stewardship)、团队绩效域(Team Performance Domain)

Learning Objectives#

Knowledge#

1	2	3	4
AI Tools Position Understand the correct positioning of AI tools in product development—augmenting human judgment, not replacing it	C-R-E-F Framework Identify the four core elements of a high-quality Prompt (C-R-E-F framework)	Sandwich Method Explain the three layers of the Sandwich Method and the corresponding responsibility boundaries	AI Red Line List at least two non-negotiable red lines in AI usage and their professional rationale

1. C-R-E-F Framework	2. Critically Review	3. Task judgement
Be able to use the C-R-E-F framework to transform vague instructions into high-quality Prompts	Be able to critically review AI-generated content and identify hallucinations and factual errors	Be able to identify which tasks are suitable for AI assistance and which require independent judgment in a product development context

Application#

- Sprint 3 Planning	- AI Red Line	- Working Habit
Apply the C-R-E-F framework in Sprint 3 Planning to use AI for User Story decomposition, then independently verify the output	Draft a simple AI red lines checklist for their own team	Establish the habit of “think independently first, then use AI to expand/verify”—not “ask AI first, accept uncritically”

You Are the Commander, Not the Passenger#

Core Metaphor#

AI is not your “all-knowing mentor.” It is more like a newly hired intern: highly intelligent, but lacking practical judgment and common sense.

PMBOK Perspective#

You are the Project Manager.
AI is the Execution Team.
Key Point: If the output is poor, the responsibility lies with the manager. It indicates that the instructions were unclear, rather than a failure on the part of the “intern.”

“Automation is for robots; Augmentation is for humans. Don’t let the tool replace your judgment.”

The commander analogy is key: a pilot who relies entirely on autopilot loses the skills to take over when it matters most. Your CS fundamentals are your manual override.

核心隐喻: AI 不是你的“大神导师”，它是你刚招聘进来的“高智商但缺乏常识的实习生”。

PMBOK 视角:

你是项目经理 (Project Manager)。

AI 是执行团队 (Execution Team)。

关键点: 如果输出结果很烂，那是管理者的责任（你的指令不明确），而不是实习生的错。

Core Skill: Prompt Engineering IS Software Engineering#

Principle: Treat natural language as a high-level programming language. Bad Prompt = Buggy Code

C — Context	R — Request	E — Examples	F — Format
“As a senior Java architect…” (role specification)“	“Explain the dependency injection pattern…”	“Refer to the implementation approach used in the Spring Framework…” (few-shot prompting)	“Compare the advantages and disadvantages using a Markdown table and include a code example.”

C-R-E-F 框架 (快速演示):

C (Context - 上下文): “作为一名资深 Java 架构师…” (设定角色)

R (Request - 任务): “解释依赖注入模式…”

E (Examples - 样本): “参考 Spring Framework 的实现方式…” (Few-shot prompting)

F (Format - 格式): “用 Markdown 表格对比优缺点，并提供一段代码示例。”

Live Demo#

Scenario A: Product Development

Bad Prompt: “Help me write a User Story.” →Result: A generic template that has little connection to the actual project.

Good Prompt (Using the C-R-E-F Framework):

C (Context): “Our product is a campus second-hand marketplace. The target users are university students, and we are currently planning Sprint 3.”
R (Request): “Please break down the following feature requirement into three independent User Stories: ‘Users can post second-hand items and set a price.’”
E (Examples): “Follow the existing format: As a [user type], I want to [action], so that [benefit].”
F (Format): “Each User Story should include: Title, Story (one sentence), Acceptance Criteria (at least three testable criteria), and a Story Points estimate (1/2/3/5/8).”

Result: Structured User Stories that can be placed directly into the Product Backlog.

Scenario B: Software Development (Additional Demonstration)

Bad Prompt: “Help me write a Python web crawler.”

Result: Generic code that may not run or fit the actual target website.

Good Prompt: Apply the C-R-E-F framework and add constraints such as “exception handling,” “compliance with the robots.txt protocol,” and “the structure of the target website.”

Result: A production-ready code snippet.

软件开发（补充演示）

Bad Prompt: “帮我写个 Python 爬虫。” → 结果：通用的、可能无法运行的代码。

Good Prompt: 套用 C-R-E-F，增加”异常处理”、”遵守 robots.txt 协议”、”目标网站结构”等约束。 → 结果：生产级别的代码片段。

Class Discussion (2 minutes): “What do these two examples have in common?”#

The quality of the prompt determines the quality of the output.

The more vague your requirements are, the less useful the AI’s response will be.

Process Management: Implementing The Sandwich Method#

Problem to Address: How can you leverage AI to improve efficiency without allowing your thinking ability to deteriorate?

🍞 Bottom Layer of the Sandwich (Human): Review, Debugging, and Integration (Quality Control)

Key Step: Always assume the AI may be hallucinating. You must run the code, and you must verify the facts.

🍖 Middle Layer of the Sandwich (AI): Heavy Coding and Draft Generation

Use AI to write boilerplate code, generate test cases, and convert data formats.

🍞 Top Layer of the Sandwich (Human): High-Level Design and Intent Definition

You need to understand system architecture and algorithmic complexity in order to define meaningful requirements for AI.

PMBOK Perspective: This corresponds to User Acceptance Testing (UAT). Before AI-generated code is merged into the main branch, it must pass your manual review.

痛点解决: 如何在利用 AI 提效的同时，不让大脑退化？
上层面包 (Human): 顶层设计与意图定义。

你需要懂架构、懂算法复杂度，才能向 AI 提需求。

中间肉饼 (AI): 繁重的编码与草稿生成。

让 AI 写样板代码 (Boilerplate)、生成测试用例、转换数据格式。

下层面包 (Human): 审查、调试与集成 (Quality Control)。

关键步骤: 永远假设 AI 在“产生幻觉” (Hallucination)。必须运行代码，必须核查事实。

Risk Control and Summary#

The Red Lines#

🔴 Data Privacy	🔴 Intellectual Property (IP)
Never paste API keys, server passwords, or proprietary company algorithms into public AI chat interfaces.	Be cautious of potential license conflicts in AI-generated code.

The Takeaway#

In the AI era, the core competitive advantage is no longer “how to write every line of code,” but “how to recognize which line of code is wrong.”
Maintain strong respect for computer science fundamentals. These foundations are what give you, as the “pilot,” the confidence and ability to take control when necessary.

红线规则 (The Red Lines):
数据隐私: 严禁将 API Keys、服务器密码或公司专有算法粘贴到公共 AI 对话框中。
知识产权: 警惕生成的代码可能存在的许可证冲突。
结语 (The Takeaway):
AI 时代的核心竞争力不再是 “如何写出每一行代码”，而是 “如何判断哪一行代码是错的”。
保持对基础知识 (CS Fundamentals) 的敬畏，这是你作为”飞行员”能够接管飞机的底气。

PMBOK Connection#

PMBOK Principles / Performance Domains	Connection to This Week’s Content
Stewardship	Professional Responsibility for AI-Generated Work: You are accountable for the quality and correctness of any work you submit, whether it was written by you or generated by AI.
Team Performance Domain	AI is changing how teams collaborate: it functions as the “zeroth team member,” but it cannot replace human communication and judgment.
Measurement Performance Domain	When using AI to generate progress reports or analyze data, the results must be verified manually. Numbers are not automatically correct simply because they were produced by AI.
Tailoring	Recognizing which tasks are suitable for AI assistance (such as boilerplate code or draft generation) and which require human judgment (such as prioritization decisions or architectural choices) is itself a professional capability in project tailoring.

“The PMBOK Stewardship principle requires project professionals to be accountable for their deliverables. ‘AI wrote it so it’s not my fault’ is not acceptable in a professional context.”

尽职管理（Stewardship）
对 AI 生成物负有专业责任——你提交的内容，无论是 AI 写的还是你写的，你都要为其质量和正确性负责
团队绩效域（Team Performance Domain）
AI 改变了团队协作方式：它是”第零个团队成员”，但不能替代人与人之间的沟通和判断
测量绩效域（Measurement Performance Domain）
用 AI 生成进度报告或分析数据时，必须对输出结果进行人工验证，数字不因为是 AI 给出的就自动正确
裁剪（Tailoring）
识别哪些任务适合 AI 辅助（样板代码、草稿生成）、哪些需要人工判断（优先级决策、架构选型），本身就是一种专业的项目裁剪能力
“PMBOK 的尽职管理原则要求项目专业人员对自己的工作成果负责。AI 不是免责声明——‘这是 AI 写的所以有错不怪我’在专业环境中不成立。”

Week 2 Seminar：Sprint 3 Planning Meeting#

1. Use C-R-E-F Framework	2. Use the top layer of the sandwich model.	3. Set a clear red line.
Ask AI to help you break down a large feature into User Stories, but you are the one who decides whether the decomposition is correct, and you are the one who writes the Acceptance Criteria.	First discuss and define the Sprint Goal independently, then use AI to validate or supplement it. Do not let AI determine the Sprint Goal for you.	Never paste customer data, API keys, or any sensitive information into public AI chat interfaces.

Week 3#

“Next week we will study Product Development Metrics—Velocity, Technical Debt, and the Burndown Chart. The data you accumulate in Sprint 3 will serve as the raw material for next week’s analysis.”

音乐

音乐

Learning and Development in the AI Era – Navigating Your Intelligent “Co-pilot”#

Learning Objectives#

Knowledge#

AI Tools Position

C-R-E-F Framework

Sandwich Method

AI Red Line

Application#

You Are the Commander, Not the Passenger#

Core Metaphor#

PMBOK Perspective#

Core Skill: Prompt Engineering IS Software Engineering#

Live Demo#

Class Discussion (2 minutes): “What do these two examples have in common?”#

Process Management: Implementing The Sandwich Method#

Risk Control and Summary#

The Red Lines#

The Takeaway#

PMBOK Connection#

Week 2 Seminar：Sprint 3 Planning Meeting#

Week 3#

支持与分享

评论区

音乐

目录