Purposeful prompting
Use PCAFE-V to connect purpose, context, audience, format, evaluation, and verification.
Module overview
Module 2 develops practical skills in structured prompting, context engineering, iterative refinement, and human-AI collaboration. You will learn to define a task before prompting, provide only necessary context, specify audience and output requirements, establish evaluation and verification criteria, decompose complex work into stages, control sources, and allocate responsibilities among users, AI systems, specialist reviewers, and final decision-makers.
Controlled workflows
Use SCOPE to organize sources, context, outputs, process stages, and evaluation gates.
Accountable collaboration
Use PARTNER to allocate tasks, authority, review, escalation, and stopping.
Module principle: A stronger prompt may improve relevance, but it does not make the output automatically accurate, suitable, authorized, or accountable.
Module learning outcomes
By the end of Module 2, you should be able to:
- Distinguish a task, prompt, context, source, output, evaluation criterion, and verification requirement.
- Design a structured prompt using the author-approved PCAFE-V framework.
- Identify assumptions, ambiguity, missing information, and unnecessary sensitive context.
- Compare initial and revised prompts using observable criteria rather than preference alone.
- Decompose a complex task into bounded stages, sources, outputs, review gates, and decisions.
- Apply the author-approved SCOPE framework to a staged AI-assisted workflow.
- Classify AI roles as tool, assistant, critic, simulator, recommender, or agent.
- Allocate tasks and authority using the author-approved PARTNER framework.
- Recognize automation bias, nominal review, excessive delegation, and responsibility laundering.
- Establish human approval, escalation, fallback, and stop conditions.
Module 2 units
Prompt Design Fundamentals
Define purpose, audience, context, format, criteria, assumptions, and verification through PCAFE-V.
Advanced Prompting and Context Engineering
Decompose tasks, manage source hierarchies, minimize context, compare iterations, and design staged workflows with SCOPE.
Human-AI Collaboration
Allocate appropriate roles, preserve meaningful oversight, and apply PARTNER to authority, review, escalation, and stopping.
Implementation note: replace placeholder anchors with the corresponding Moodle activity or section URLs.
Embedded Module 2 study guide
Use this two-week plan unless the official Moodle schedule indicates otherwise.
Week 3: Unit 4 and the beginning of Unit 5
Recommended: 8–10 hours- Review Module 1.
Retrieve the AI Paradox, ORBIT, fluency-reliability distinction, and human-accountability principle. - Read Chapter 4 and complete Unit 4.
Diagnose weak prompts and build complete PCAFE-V prompts. - Compare prompt versions.
Evaluate outputs against purpose, relevance, structure, evidence, and verification needs. - Practise context minimization.
Separate necessary, optional, sensitive, and prohibited information. - Begin Chapter 5.
Study decomposition, context engineering, source hierarchies, and prompt sequences. - Start Assignment 2.
Select portfolio tasks, create a prompt log, and preserve initial and revised versions.
Evidence to retain
- Weak-prompt diagnosis
- PCAFE-V redesign
- Initial and revised outputs
- Evaluation notes
- Privacy-preserving context revision
- Assignment 1 reflection notes
Week 4: Complete Units 5 and 6
Recommended: 8–10 hours- Complete Unit 5.
Build a SCOPE sequence with bounded stages, source controls, outputs, and decision gates. - Read Chapter 6 and complete Unit 6.
Compare human and AI strengths without assigning the AI authority it cannot hold. - Build a responsibility matrix.
Assign tasks to the user, AI, specialist reviewer, approver, and stop authority. - Audit automation bias.
Identify anchoring, rubber-stamping, nominal review, and responsibility laundering. - Complete Assignment 2.
Review portfolio evidence, explanations, verification, disclosure, and file requirements. - Consolidate and transition.
Complete the module check and prepare for ethics, fairness, privacy, and integrity in Module 3.
Evidence to retain
- SCOPE workflow
- PARTNER responsibility matrix
- Source and context boundary
- Review, escalation, and stop gates
- Assignment 2 final self-review
- Updated Assignment 1 reflection
Show the recommended prompt-development cycle
- Define the human purpose and task.
- Make an independent initial attempt where learning or judgment requires it.
- Design the prompt with PCAFE-V.
- Limit context and sources.
- Generate a provisional output.
- Evaluate it against explicit criteria.
- Verify consequential claims independently.
- Revise the prompt or workflow.
- Document human decisions and remaining uncertainty.
Assessment milestones in Module 2
Assignment 2: Prompt Engineering Portfolio
Weight: 15%
Start: During Unit 4, Week 3
Develop: Across Units 4–6
Recommended submission: After Module 2, at the end of Week 4
Official Moodle due date and time: ____________________
Assignment 1: Journal continues
Record what changed when prompts became more structured, what context was removed, how outputs were evaluated, where human review mattered, and how automation bias affected judgment.
Recommended final submission: End of Module 6, Week 11
Official Moodle due date and time: ____________________
Begin Assignment 2 early: The portfolio should demonstrate a development process. Preserve initial prompts and outputs rather than showing only polished final versions.
Show the Assignment 2 working-record checklist
- Task and legitimate purpose
- Initial prompt and output
- PCAFE-V diagnosis and redesign
- Context and source decisions
- Evaluation criteria
- Verification performed
- Revision rationale
- Final human decision
- AI-use disclosure required by the assignment
Core learning activities
| Activity | Purpose | Product to retain | Connection |
|---|---|---|---|
| Weak-prompt diagnosis | Identify missing purpose, context, audience, format, criteria, and verification. | Annotated diagnosis | Assignment 2 |
| PCAFE-V redesign | Transform an unstructured request into a purposeful prompt. | Initial and revised prompt | Assignments 1 and 2 |
| Prompt comparison | Evaluate output differences using observable criteria. | Comparison record | Assignment 2 |
| Context minimization | Remove unnecessary identifying or sensitive detail. | Context boundary | Assignments 2, 3, 4, and 6 |
| SCOPE sequence | Decompose a complex task into stages and decision gates. | Workflow map | Assignments 4 and 6 |
| PARTNER matrix | Allocate tasks, review, approval, and stop authority. | Responsibility matrix | Assignments 3, 4, and 6 |
| Automation-bias scenario | Recognize anchoring and nominal review. | Risk and safeguard notes | Assignments 1, 3, and 4 |
Live AI and non-platform options
Option A: Approved AI
Use an approved system with synthetic, public, supplied, or authorized information.
Option B: Supplied materials
Use supplied prompts, contexts, outputs, and revision records.
Option C: Non-AI process
Design and evaluate equivalent task instructions and workflows without operating an AI platform.
All options must assess the same prompt-design, workflow, evaluation, and responsibility outcomes.
Frameworks introduced in Module 2
PCAFE-VSCOPEPARTNERHuman oversightSource hierarchyReview gatesStop conditions
PCAFE-V
Use the exact author-approved expansion supplied in the textbook and course materials. Apply it to prompt purpose, context, audience, format, evaluation, and verification.
SCOPE
Use the author-approved framework to structure sources, context, outputs, process stages, and evaluation or decision gates.
PARTNER
Use the author-approved framework to allocate human and AI roles, authority, review, escalation, and stopping.
Responsible prompting, accessibility, and support
Boundaries
- Enter only information necessary and authorized for the task.
- Do not treat a detailed prompt as proof of a trustworthy output.
- Do not assign final approval or accountability to an AI system.
- Do not enable external actions for course activities.
- Verify important claims and document material assistance.
Support
- Use supplied alternatives if live AI is inaccessible, unavailable, or unwanted.
- Contact the course coordinator for assignment and permitted-use questions.
- Use library support for sources and verification.
- Contact accessibility or technical services when needed.