Table of Contents
Overview
This page provides the cumulative consolidation materials for Unit 4. It includes a consolidation exercise, a cumulative homework set, a cumulative project, a cumulative self‑check quiz, and guides for Assignment 2 and Assignment 1. Complete these activities after finishing Lessons 4.1–4.4 to solidify your understanding and prepare for Unit 5.
The consolidation components are formative and ungraded (unless your instructor designates otherwise). They are intended to help you solidify your understanding, identify gaps, and build confidence before moving on to advanced prompting and context engineering in Unit 5.
A. Unit 4 Consolidation Exercise
- Estimated time: 45–60 minutes.
- Instructions: Read the ambiguous, over‑detailed, under‑specified prompt below. Then complete the 15 tasks that follow. Use the provided response template to structure your work. After completing, expand the details for guided feedback and a solution design.
Use AI to help me with my project. I need something about data privacy. My team is working on this for our company. We are a tech startup in San Francisco. Our CEO wants a report. Make it professional and include some examples. We need it by next week. Also, I want it to be comprehensive and accurate. Maybe include some statistics. We have a lot of data, but I can't share it because it's private. Just use what you know.
Tasks:
- Identify the human problem.
- Define a legitimate purpose.
- Distinguish purpose, task, output, intended use, and final human action.
- Identify ambiguity and hidden assumptions.
- Define scope and exclusions.
- Identify necessary context.
- Remove unnecessary or sensitive context.
- Define the audience.
- Select a suitable output format.
- Write observable evaluation criteria.
- Add verification requirements.
- Include missing‑information and uncertainty rules.
- Produce a revised PCAFE‑V prompt.
- Identify remaining AI limitations.
- State required human review.
Response template: Write your answers in a document. For each task, use a clear heading (e.g., "1. Human Problem") and provide your analysis. After you finish, check your work against the guidance below.
Show guided feedback and solution design
1. Human problem: The team needs a report on data privacy to inform decision‑making, but the request is vague and lacks specifics.
2. Legitimate purpose: "Prepare a clear, evidence‑based overview of current data privacy challenges and best practices for our tech startup, so that our leadership team can make informed decisions about our privacy strategy."
3. Distinctions:
- Purpose: Inform leadership about data privacy to guide strategy.
- AI task: Generate a draft report synthesizing publicly available information.
- Output: A structured briefing note.
- Intended use: Leadership will review the draft and use it as a discussion starter.
- Final human action: Leaders will decide on privacy strategy based on the review.
4. Ambiguities & assumptions: "Professional," "comprehensive," "accurate" are undefined. Assumes AI knows the company's context, has current data, and can access private information.
5. Scope & exclusions: Scope: public information only, general best practices. Exclusions: internal data, financial projections, legal advice.
6. Necessary context: Type of company (tech startup), need for privacy best practices, intended audience (leadership).
7. Removed context: Location (San Francisco), CEO's name, deadline (can be mentioned outside the prompt), internal data.
8. Audience: Leadership team members with varying technical backgrounds; need clear, actionable insights.
9. Format: One‑page briefing note with sections: Overview, Key Privacy Challenges, Best Practices, Recommended Actions.
10. Evaluation criteria: Relevance, clarity, completeness, source support, concision, uncertainty.
11. Verification requirements: All claims must be traceable to public sources; cite sources; mark unverified claims; no fabrication.
12. Missing‑information & uncertainty rules: Indicate when data is incomplete; use tentative language; present conflicting views.
13. Revised prompt: (See example below.)
14. Remaining limitations: AI may not have the latest regulations; cannot replace legal advice; may miss nuance.
15. Human review: Leadership must verify all claims, check sources, and consult legal counsel before making decisions.
Purpose: Prepare a clear, evidence‑based overview of current data privacy challenges and best practices for our tech startup's leadership team, to inform strategic decisions.
Context: Use only publicly available information from reputable sources (e.g., government publications, academic journals, industry reports). Do not use internal data.
Audience: Leadership team members with varied technical backgrounds. They need clear, actionable insights without excessive jargon.
Format: One‑page briefing note with sections:
1. Overview of data privacy landscape
2. Key challenges (bullet points)
3. Best practices (bullet points)
4. Recommended next steps
Length: Under 500 words. Use plain language.
Evaluation Criteria:
- Relevance: Focus on tech startups and current privacy issues.
- Clarity: Define technical terms.
- Completeness: Cover all four sections.
- Source support: Cite all claims.
- Concision: Under 500 words.
- Uncertainty: Note gaps and conflicting views.
Verification Requirements:
- All factual claims must be traceable to cited sources (include URLs or references).
- Mark unverified claims explicitly.
- Do not invent data or sources.
- If information is missing, indicate what is needed.
- Mark the briefing as a draft requiring leadership review.
This briefing is not legal advice and must be reviewed by counsel before any decisions are made.
B. Unit 4 Cumulative Homework Set
This set of 14 problems covers the entire unit. They are formative and ungraded, and are excellent preparation for Assignment 2 and Unit 5. Complete them after finishing all lessons and the consolidation exercise.
Instructions: Define prompt and prompt design in your own words. Include the role of each.
Show sample solution
Prompt: An instruction given to a generative AI system to elicit a response; in responsible use, it is a structured specification that includes purpose, context, audience, format, criteria, and verification.
Prompt design: The practice of crafting intentional, structured, and responsible instructions for generative AI, with attention to the six PCAFE‑V elements.
Instructions: Why must a purpose be legitimate and human‑centred? Give an example of a purpose that is not legitimate.
Show sample solution
A legitimate purpose is human‑centred, ethically grounded, bounded, and accountable. A purpose that is not legitimate might be: "Generate fake reviews to boost our product rating." This is deceptive and unethical.
Instructions: Distinguish the human purpose from the AI task in the following: "Help me draft a letter to request a meeting with a potential client."
Show sample solution
Human purpose: Secure a meeting with a potential client to discuss a business opportunity.
AI task: Generate a draft letter that can be reviewed and sent.
Instructions: In the previous example, what is the intended use and the final human action?
Show sample solution
Intended use: Review the draft, customize it, and use it to reach out to the client.
Final human action: Send the letter (or revised version) and follow up.
Instructions: Identify at least three ambiguous terms in the prompt: "Write a compelling case study about our successful project."
Show sample solution
- "Compelling" — what makes it compelling?
- "Successful" — how is success measured?
- "Our project" — which project?
Instructions: What hidden assumptions are present in: "Summarize the latest trends in your field."
Show sample solution
- The AI knows what "your field" is.
- The AI has access to the latest trends.
- The AI can distinguish trends from fads.
Instructions: Define an appropriate scope for a prompt that asks the AI to "propose improvements to a customer service process."
Show sample solution
"Propose three improvements to our customer service response time, based on best practices in the retail sector. Do not include specific employee names or internal performance data. Focus on system‑level changes."
Instructions: Which of these details is most relevant to a prompt about "drafting a social media post for a product launch"? (a) Company history, (b) Product features, (c) Office location, (d) CEO's hobbies.
Show answer
Answer: (b) Product features.
Instructions: Minimize the context in: "I am a teacher at Lincoln High School in Chicago. I have 30 students in my AP Biology class. I need a worksheet on mitosis. The school is closed next Monday."
Show sample solution
"Create a worksheet on mitosis for AP Biology students." (Remove location, class size, school closure.)
Instructions: Classify the following: "A list of employee social insurance numbers."
Show answer
Personal and Confidential. Should never be shared with an AI system.
Instructions: Write an audience profile for a prompt that asks the AI to "explain blockchain to a group of senior executives with no technical background."
Show sample solution
- Prior knowledge: None or very little.
- Language: Plain English, avoid jargon.
- Purpose: To understand blockchain's potential business impact.
- Tone: Professional but accessible.
- Action: They will decide whether to invest in blockchain.
Instructions: Add an accessibility criterion to the prompt: "Create a presentation on renewable energy."
Show sample solution
"Ensure the presentation is accessible: use sans‑serif fonts, high contrast, alt text for images, and avoid colour‑only cues. Provide a text transcript."
Instructions: Match the purpose to the best format: (a) comparing products, (b) step‑by‑step procedure, (c) brainstorming ideas.
Show answer
- (a) Table
- (b) Checklist or step‑by‑step guide
- (c) Question set or outline
Instructions: Write verification requirements for a prompt that asks the AI to "provide a summary of a clinical trial."
Show sample solution
- All data must be traceable to the trial protocol or published results.
- Include study identifiers and date.
- Clearly distinguish between primary and secondary outcomes.
- Mark any unverified claims.
- Do not interpret efficacy beyond the data.
- Mark as a draft requiring medical review.
After completing, use the self‑check criteria: Did you address each element of PCAFE‑V? Are your answers specific and observable?
C. Unit 4 Cumulative Project
- Estimated effort: 3–4 hours.
Context choices: Select one bounded, low‑ or moderate‑risk task from:
- Education or learning
- Professional practice
- Organizational activity
- Communication
- Creative work
- Public information
- Everyday life
Deliverable: Create a document with the following 38 sections (suggested length: 1,250–1,500 words or equivalent structured submission).
- Use‑case title
- Human problem
- Legitimate purpose
- Intended outcome
- AI‑supported task
- Requested output
- Intended use
- Final human action
- Non‑AI alternative
- Scope
- Exclusions
- Initial prompt
- Initial output (or supplied)
- PCAFE‑V diagnosis
- Necessary context
- Excluded context
- Information classification
- Audience profile
- Output format
- Evaluation criteria
- Verification requirements
- Missing‑information rule
- Uncertainty rule
- Conflict‑handling rule
- No‑fabrication rule
- Revised PCAFE‑V prompt
- Revised output (or supplied)
- Side‑by‑side evaluation (table)
- Improvements
- Regressions or new problems
- Remaining limitations
- Claims requiring verification
- Human‑review requirement
- Final human assessment
- Prompt‑version record
- Material AI‑use disclosure
- Assignment 1 reflection connection
- Preparation for Unit 5
Self‑review checklist:
- Is the purpose legitimate and human‑centred?
- Are context, audience, format, criteria, and verification all specified?
- Is the revised prompt a clear improvement over the initial?
- Are remaining limitations and verification needs explicitly stated?
- Is the human review plan realistic and documented?
Show staged hints
Hint 1: Build on the mini‑projects from Lessons 4.1–4.4. You already have most of the pieces.
Hint 2: The evaluation table (section 28) is key. Use common criteria to compare initial and revised outputs.
Hint 3: Be honest about limitations. No prompt is perfect.
Hint 4: The documentation record (section 35) should include version numbers, dates, and decisions.
Show solution architecture
- Start with a clear, bounded use case.
- Write the initial prompt as you would have before Unit 4.
- Generate or obtain an initial output.
- Systematically apply PCAFE‑V to diagnose weaknesses.
- Redesign the prompt with all six elements.
- Generate the revised output.
- Compare outputs using a table with your criteria.
- Document improvements, regressions, and remaining issues.
- Specify verification needs and a human review plan.
- Write your final assessment and disclosure.
Show illustrative outline (different context)
Use‑case: Planning a community garden project.
Purpose: Help me prepare a one‑page proposal to present to the neighbourhood association.
Context: Use local zoning bylaws and best practices from similar projects. Exclude personal details.
Audience: Neighbourhood association board members; they need clear benefits and feasibility.
Format: Proposal with sections: Overview, Benefits, Site Plan, Budget Estimate, Next Steps.
Criteria: Relevance, clarity, completeness, feasibility, actionability.
Verification: All data must be traceable to city documents; mark cost estimates as approximate; note any zoning uncertainties.
Revised prompt: (full text incorporating above).
Comparison: Initial output was vague; revised output is structured, evidence‑based, and transparent about gaps.
Limitations: Still requires site‑specific verification and board feedback.
D. Unit 4 Cumulative Self‑Check Quiz
Test your overall understanding of Unit 4. Answer each question, then expand the details to check your work.
Q1: What are the six elements of PCAFE‑V?
Show answer
Correct answer: a) Purpose, Context, Audience, Format, Evaluation, Verification.
Explanation: PCAFE‑V is the framework taught throughout Unit 4.
Q2: Why should a prompt begin with a legitimate human purpose?
Show answer
Correct answer: b) To ensure the output serves a real need and remains accountable.
Q3: Which of the following is an appropriate exclusion for a prompt?
Show answer
Correct answer: b) "Do not include financial projections." This is a clear, specific exclusion.
Q4: What is the difference between context and audience?
Show answer
Correct answer: a) Context is the information and constraints; audience is the intended user or reader.
Q5: Why is "make it professional" insufficient as an evaluation criterion?
Show answer
Correct answer: a) It is not specific or observable; it is a value judgment.
Q6: What does the no‑fabrication rule prevent?
Show answer
Correct answer: b) It prevents the AI from inventing citations, data, or facts.
Q7: Why is "verify your answer" not sufficient for verification?
Show answer
Correct answer: d) All of the above. The AI cannot verify itself, it is vague, and it lacks independent access.
Q8: What is the final step in the prompt refinement cycle?
Show answer
Correct answer: d) Record the human decision.
Q9: Which of the following is a responsible‑use consideration?
Show answer
Correct answer: c) Preserve human responsibility for evaluation and decisions.
Q10: What is the role of format in a prompt?
Show answer
Correct answer: b) To make the output usable and reviewable.
Q11: When handling conflicting evidence, what should the AI do?
Show answer
Correct answer: b) Present both perspectives and note the conflict.
Q12: What is the purpose of documenting prompts and outputs?
Show answer
Correct answer: d) All of the above.
Q13: Which of the following is NOT a characteristic of well‑designed context?
Show answer
Correct answer: c) Excessive. Context should be minimized, not excessive.
Q14: What is the danger of a polished but substantively weak output?
Show answer
Correct answer: b) It may appear trustworthy and lead to over‑reliance.
Q15: Which of the following is an example of an observable evaluation criterion?
Show answer
Correct answer: b) "Include at least three supporting references." This is observable and checkable.
E. Assignment 2 Portfolio Starter Guide
Assignment 2: Prompt Engineering Portfolio (15%) requires you to submit three complete entries in three contexts: educational, professional/workplace, and personal/everyday life.
For each entry, retain:
- Legitimate purpose
- Initial prompt and initial output
- PCAFE‑V diagnosis
- Revised prompt and revised output
- Common evaluation criteria comparison
- Verification needs and remaining limitations
- Human decisions and AI‑use disclosure
Suggested organisation:
- Use a consistent naming convention (e.g., Entry1_initial_prompt.txt).
- Keep a version log.
- Use a comparison table (like the one in the worked example).
- Include a final reflection on what you learned.
Do not use real personal, confidential, or proprietary data. Use synthetic or public information only. Follow your institution's AI‑use policy.
Show self‑review checklist
- Is the purpose clearly stated?
- Are all six PCAFE‑V elements present?
- Is the initial output clearly weaker than the revised?
- Are verification needs explicitly listed?
- Is the human decision documented?
- Is there a transparency disclosure?
F. Assignment 1 Journal Guide
For your Assignment 1: AI Learning and Reflection Journal, consider these prompts that span the entire unit. Choose 3–5 to reflect on in detail.
- How did defining a legitimate human purpose change the way you approached prompting?
- What hidden assumptions did you discover in your own prompts, and how did you address them?
- How did minimizing context improve your prompts? What did you remove?
- How did defining the audience change the tone and content of your outputs?
- What format choices did you make, and why? Did they improve reviewability?
- How did converting vague instructions into observable criteria help you evaluate outputs?
- What verification requirements did you add, and how did they make the output more trustworthy?
- What limitations remained despite your best prompting efforts? What does that tell you about human responsibility?
- How has your understanding of "responsible AI use" evolved?
- What will you do differently in your future prompting practice?
Show reflection tips
- Be specific: reference actual prompts and outputs.
- Connect to PCAFE‑V elements.
- Acknowledge challenges and how you worked through them.
- Identify remaining questions for further exploration.
G. Key Takeaways from Unit 4
- Purpose is the foundation of every responsible prompt.
- Context must be minimized and relevant.
- Audience must be defined with specificity.
- Format makes outputs usable and reviewable.
- Evaluation criteria must be observable and consistent.
- Verification is a human responsibility; ask for traceability and mark uncertainty.
- Prompt refinement is an iterative cycle; compare versions using common criteria.
- Even the best prompt cannot eliminate the need for human judgment and accountability.
- Document your work to support transparency and learning.
Looking Ahead to Unit 5
Congratulations on completing Unit 4! You have built a strong foundation in prompt design fundamentals: purpose, context, audience, format, evaluation criteria, and verification. You can now construct complete PCAFE‑V prompts, diagnose weaknesses, refine prompts purposefully, and identify remaining limitations.
In Unit 5: Advanced Prompting and Context Engineering, you will extend these skills to more complex workflows:
- Advanced context engineering: Managing large amounts of context, source hierarchies, and context drift.
- Task decomposition: Breaking complex tasks into smaller, manageable steps.
- Prompt sequences: Using multiple prompts in a chain to achieve a complex outcome.
- Multi-stage workflows: Coordinating AI and human activities across stages.
- SCOPE application: A framework for managing source-controlled, context-rich prompts.
Unit 5 will build directly on the foundations you have established in Unit 4. You will also continue work on Assignment 2: Prompt Engineering Portfolio, which will be submitted at the end of Module 2.
- Complete the consolidation exercise and cumulative project.
- Draft your Unit 4 journal entries.
- Review your portfolio entries against the Assignment 2 checklist.
- Begin reading Unit 5 materials.