Unit 5 Consolidation

2. Part A — Unit 5 Consolidation Exercise

The Request

You are a policy analyst in a mid-sized organisation. A senior director has sent you the following message:

Your Task

Transform this over-broad request into a controlled, SCOPE-based workflow. Complete each of the twenty-one tasks below. Write your response in the response template provided.

1. Define the purpose.
2. Restrict the scope.
3. Select the system.
4. Classify the sources.
5. Build a source hierarchy.
6. Define a conflict rule.
7. Minimise and structure the context.
8. Separate instructions from evidence.
9. Decompose the task.
10. Build a prompt sequence.
11. Specify structured inputs.
12. Specify structured outputs.
13. Add verification points.
14. Add human review.
15. Define output permissions.
16. Identify drift indicators.
17. Address the untrusted instruction.
18. Add escalation.
19. Add stop conditions.
20. Apply SCOPE.
21. Identify remaining limitations.

Response Template

=== UNIT 5 CONSOLIDATION EXERCISE – RESPONSE ===

1. Purpose:

2. Scope (included / excluded):

3. System:

4. Source classification:
   – Source A:
   – Source B:
   – Source C:
   – Source D:

5. Source hierarchy:
   1.
   2.
   3.
   4.

6. Conflict rule:

7. Minimised and structured context:

8. Instruction–evidence boundary:

9. Task decomposition (stages):

10. Prompt sequence (stage prompts):

11. Structured inputs:

12. Structured outputs:

13. Verification points:

14. Human review points:

15. Output permissions:

16. Drift indicators:

17. Handling of untrusted instruction:

18. Escalation conditions:

19. Stop conditions:

20. SCOPE assessment:

21. Remaining limitations:

Self-Assessment Checklist

• Did I define a clear, bounded purpose?
• Did I restrict the scope and identify exclusions?
• Did I classify all sources?
• Did I build a source hierarchy with a conflict rule?
• Did I minimise the context?
• Did I separate instructions from evidence?
• Did I decompose the task into logical stages?
• Did I write a prompt for each stage?
• Did I specify structured inputs and outputs?
• Did I add verification and human review points?
• Did I define output permissions?
• Did I identify drift indicators?
• Did I address the untrusted instruction?
• Did I add escalation and stop conditions?
• Did I apply all five SCOPE elements?
• Did I identify remaining limitations?

Staged Hints

Solution Design

Illustrative Partial Example (Different Context)

1. Purpose: To create a 5-page study guide for the final exam.
2. Scope: Included: syllabus, lecture notes, textbook chapters.
   Excluded: personal notes, external sources.
3. System: Approved AI tool.
4. Source classification: Syllabus (authoritative), lecture notes
   (supporting), textbook (authoritative).
5. Source hierarchy: Syllabus, textbook, lecture notes.
6. Conflict rule: Syllabus prevails.
7. Minimised context: Extract relevant sections.
8. Instruction–evidence boundary: Instructions separate from
   source content.
9. Stages: Topic extraction, definition gathering, synthesis,
   review.
10. Prompt sequence: PCAFE-V for each stage.
11. Structured inputs: Task info, source records.
12. Structured outputs: Topic list with source references.
13. Verification: After each stage.
14. Human review: The student reviews the guide.
15. Output permissions: Draft = personal study. Final = share with
    study group.
16. Drift indicators: Scope expansion (adding unrelated topics).
17. Untrusted instruction: None identified.
18. Escalation: If a topic is unclear, ask the instructor.
19. Stop conditions: If the syllabus is unavailable, stop.
20. SCOPE assessment: All five elements applied.
21. Remaining limitations: The guide does not replace attending
    lectures.

3. Part B — Unit 5 Cumulative Homework Set

Estimated time: 2–2.5 hours · Formative · Ungraded · Recommended portfolio preparation.

This homework set contains eighteen problems covering all Unit 5 topics. Work through them in order. Answers, solution designs, and hints are provided in the expandable details.

Purpose: To summarise each authoritative source.
Context: The sources classified as authoritative in Stage 1.
Audience: The analyst who will synthesise the findings.
Format: Numbered list with source title, 100-word summary, and
  source date.
Evaluation: Each summary accurately reflects the source.
Verification: Compare each summary to the original source.
Prompt:
"Using the attached authoritative sources, provide a 100-word
summary of each. For each source, include the title, the summary,
and the date. Do not add interpretation. Output: a numbered
list."

=== TASK INFORMATION ===
Purpose: Summarise the policy for internal staff.
Audience: Staff members.
Format: 150-word summary.

=== SOURCE RECORDS ===
Source 1 – Title: [Title], Authority: [Issuer], Date: [Date],
  Status: Authoritative.

=== CONSTRAINTS ===
– Do not exceed 150 words.
– Use plain language.

=== REQUIRED OUTPUT ===
A 150-word summary with a claim inventory.

=== RISK REGISTER ===
Risk 1: [Description]
  – Likelihood: [High/Medium/Low]
  – Impact: [High/Medium/Low]
  – Source: [Reference]
  – Mitigation: [Description]

=== UNCERTAINTY LOG ===
– [Uncertainty 1]

Draft: Internal review only. Not for external sharing or
  publication.
Revised: May be shared with the policy team.
Final: May be shared with the executive team after approval.
Prohibited: Not for legal advice or external advocacy.

4. Part C — Unit 5 Cumulative Project

Available Contexts

Select one bounded, low- or moderate-risk task from:

• Learning: Preparing a study guide, exam review, or course summary.
• Professional practice: Preparing a policy summary, briefing note, or analysis.
• Communication: Drafting a public-facing summary or FAQ.
• Research support: Conducting a literature review or synthesising findings.
• Organisational activity: Preparing an internal process document.
• Public information: Summarising public policy or guidelines.
• Everyday life: Planning a complex personal project.

Required Sections

Your project must include all forty-one sections listed below.

1. Use-case title
2. Human problem
3. Legitimate purpose
4. Intended outcome
5. Authorised AI system
6. Scope
7. Exclusions
8. Information boundaries
9. Source inventory
10. Source classification
11. Source hierarchy
12. Date and version controls
13. Conflict rule
14. Missing-information rule
15. Context package
16. Instruction–evidence boundary
17. Task-decomposition map
18. Dependencies
19. Prompt sequence
20. Structured inputs
21. Structured outputs
22. Traceability requirements
23. Evaluation criteria
24. Verification points
25. Human review points
26. Specialist review (if required)
27. Provisional-output labels
28. Output permissions
29. Context-drift indicators
30. Workflow-drift indicators
31. Prompt-injection controls
32. Exception handling
33. Escalation
34. Stop conditions
35. Change and revalidation triggers
36. Complete SCOPE analysis
37. Remaining limitations
38. Final recommendation
39. Assignment 2 portfolio connection
40. Assignment 1 reflection connection
41. Preparation for Unit 6

Instructions

1. Select a bounded, low- or moderate-risk task.
2. Use synthetic, public, or generalised information only.
3. Complete each of the forty-one sections.
4. Use the response template provided.
5. Review your work using the self-review checklist.

Response Template

=== UNIT 5 CUMULATIVE PROJECT – RESPONSE ===

1. Use-case title:

2. Human problem:

3. Legitimate purpose:

4. Intended outcome:

5. Authorised AI system:

6. Scope:

7. Exclusions:

8. Information boundaries:

9. Source inventory:

10. Source classification:

11. Source hierarchy:

12. Date and version controls:

13. Conflict rule:

14. Missing-information rule:

15. Context package:

16. Instruction–evidence boundary:

17. Task-decomposition map:

18. Dependencies:

19. Prompt sequence:

20. Structured inputs:

21. Structured outputs:

22. Traceability requirements:

23. Evaluation criteria:

24. Verification points:

25. Human review points:

26. Specialist review (if required):

27. Provisional-output labels:

28. Output permissions:

29. Context-drift indicators:

30. Workflow-drift indicators:

31. Prompt-injection controls:

32. Exception handling:

33. Escalation:

34. Stop conditions:

35. Change and revalidation triggers:

36. Complete SCOPE analysis:

37. Remaining limitations:

38. Final recommendation:

39. Assignment 2 portfolio connection:

40. Assignment 1 reflection connection:

41. Preparation for Unit 6:

Assessment Criteria

Privacy and Integrity Requirements

• Do not use real personal, confidential, proprietary, or security-sensitive information.
• Use synthetic, public, or generalised information.
• If you use real sources, ensure they are publicly available and properly cited.
• Do not use this project for high-stakes decisions.
• All work you submit must be your own.

Self-Review Checklist

• Have I completed all forty-one sections?
• Have I classified all sources?
• Have I separated instructions from evidence?
• Have I designed a complete prompt sequence?
• Have I added verification and review gates?
• Have I defined permission boundaries?
• Have I identified drift indicators?
• Have I addressed prompt injection?
• Have I defined escalation and stop conditions?
• Have I identified revalidation triggers?
• Have I avoided using real confidential information?

1. Use-case title: COMP 3270 Final Exam Study Guide
2. Human problem: The course has 7 modules and 20 units; a
   concise study guide is needed.
3. Legitimate purpose: To support revision for the final exam.
4. Intended outcome: A 5-page study guide with key topics,
   definitions, and practice questions.
5. Authorised AI system: Institutionally approved AI tool.
6. Scope: Modules 1–7, key concepts, definitions, and practice
   questions.
7. Exclusions: Personal notes, unofficial summaries, external
   sources.
8. Information boundaries: No personal information; only public
   course materials.
9. Source inventory: Syllabus, lecture notes, textbook chapters.
10. Source classification: Syllabus (authoritative), lecture notes
    (supporting), textbook (authoritative).
11. Source hierarchy: Syllabus > textbook > lecture notes.
12. Date and version controls: Fall 2026 syllabus, v1.0.
13. Conflict rule: Syllabus prevails; if conflict, ask instructor.
14. Missing-information rule: Mark gaps; consult instructor.
15. Context package: Structured by module.
16. Instruction-evidence boundary: Instructions separated from
    source content.
17. Task-decomposition map: 4 stages (extract topics, gather
    definitions, synthesise, review).
18. Dependencies: Each stage depends on the previous.
19. Prompt sequence: One prompt per stage.
20. Structured inputs: Task info, source records, constraints.
21. Structured outputs: Topic list, definitions, practice
    questions.
22. Traceability requirements: Each topic linked to source.
23. Evaluation criteria: Coverage, accuracy, readability.
24. Verification points: After each stage.
25. Human review points: After synthesis and before finalisation.
26. Specialist review: Instructor if needed.
27. Provisional-output labels: "DRAFT – STUDY GUIDE."
28. Output permissions: Personal use; may share with study group.
29. Context-drift indicators: Audience shift, topic expansion.
30. Workflow-drift indicators: Skipped review.
31. Prompt-injection controls: Only approved sources;
    instruction-evidence separation.
32. Exception handling: If a topic is unclear, mark it as a gap.
33. Escalation: Ask instructor for clarification.
34. Stop conditions: If the syllabus is unavailable, stop.
35. Change and revalidation triggers: New course version, new
    instructor guidance.
36. Complete SCOPE analysis: All five elements applied.
37. Remaining limitations: Study guide does not replace lectures.
38. Final recommendation: Reuse the workflow for future exams.
39. Assignment 2 portfolio connection: Source inventory, prompt
    sequence.
40. Assignment 1 reflection connection: Reflection on source
    control.
41. Preparation for Unit 6: Consider human–AI role allocation.

5. Part D — Unit 5 Cumulative Self-Check Quiz

Estimated time: 30–40 minutes · Formative · Ungraded.

This quiz contains sixteen questions covering all of Unit 5. Answer each question, then check your response.

6. Part E — Assignment 2 Workflow Extension Guide

What to Extend

Select at least one of your three portfolio contexts and add the following elements:

• Source inventory — the sources you considered.
• Source hierarchy — priority order with conflict and missing-information rules.
• Context package — structured context with purpose, instructions, definitions, sources, constraints, gaps, conflicts, and prohibited uses.
• Task-decomposition map — stages, dependencies, inputs, and outputs.
• Prompt sequence — stage-specific prompts using PCAFE-V.
• Structured output — at least one structured output format (e.g., claim inventory, source table).
• Verification points — where and how you verified the output.
• Drift analysis — context, conversation, task, output, and workflow drift indicators.
• SCOPE analysis — a complete SCOPE assessment.
• Human decisions — where humans intervened and why.
• Remaining limitations — what the workflow cannot address.

Suggested Organisation

1. Portfolio cover page
2. Original prompt entry (from earlier Unit)
3. Context-engineered extension:
   • Source inventory
   • Source hierarchy
   • Context package
   • Task decomposition
   • Prompt sequence
   • Structured output
   • Verification points
   • Drift analysis
   • SCOPE analysis
   • Human decisions
   • Remaining limitations
4. Reflection

Evidence Checklist

• Source inventory is complete.
• Source hierarchy is documented with conflict and missing-information rules.
• Context package is structured and minimised.
• Task decomposition is logical.
• Prompt sequence uses PCAFE-V for each stage.
• At least one structured output is included.
• Verification points are documented.
• Drift analysis addresses all five types.
• SCOPE analysis is complete.
• Human decisions are recorded.
• Remaining limitations are acknowledged.

Version-Naming Convention

Portfolio entry: [Context] – [Task]
Original prompt: [Context]-prompt-v1
Context package: [Context]-context-v1
Prompt sequence: [Context]-sequence-v1
Structured output: [Context]-output-v1

Workflow-Record Template

=== WORKFLOW RECORD ===
Use case:
Purpose:
System:
Scope:
Sources:
Stages:
Prompts:
Outputs:
Verification points:
Human review points:
Permissions:
Drift indicators:
Injection controls:
Escalation:
Stop conditions:
Revalidation triggers:

Privacy Reminder

• Do not include personal, confidential, proprietary, or security-sensitive information.
• Use synthetic, public, or generalised information.
• If you use real sources, ensure they are publicly available and properly cited.

Academic-Integrity Reminder

• All work you submit must be your own.
• Do not submit work generated entirely by AI.
• Use AI as a tool for thinking and drafting, not as a substitute for your own analysis.

Self-Review Checklist

• Have I extended at least one portfolio entry?
• Have I documented the source inventory and hierarchy?
• Have I included a context package?
• Have I created a task-decomposition map?
• Have I written a prompt sequence?
• Have I included a structured output?
• Have I documented verification points?
• Have I completed a drift analysis?
• Have I completed a SCOPE analysis?
• Have I recorded human decisions and remaining limitations?

7. Part F — Assignment 1 Journal Guide

Journal Prompts

1. Context engineering vs. prompt writing: How does context engineering differ from prompt writing in your own practice? What new practices have you adopted?
2. Source authority: Which source did you treat as authoritative in a recent task, and how did you justify that decision? Were there any challenges?
3. Context minimisation: What information did you exclude from a recent context package, and why? What did you include, and why?
4. Task decomposition: How did you decide how many stages to use in a workflow? What dependencies did you identify, and how did you manage them?
5. Error propagation: Where could errors propagate in your workflow? What did you do to prevent this?
6. Structured outputs: How did structured outputs affect your review process? Did they make review easier or harder?
7. Drift: Have you observed any form of drift in your AI-supported work? What happened, and what did you do?
8. Prompt injection: How would you handle untrusted source instructions in your workflow? What controls have you implemented?
9. SCOPE: Did SCOPE improve your workflow? Which element was most useful, and why?
10. Human responsibility: Which responsibilities remained human in your workflow, and why?
11. Practice change: How has your practice changed as a result of Unit 5? What will you do differently going forward?

Reflection-Planning Checklist

• I can describe how context engineering differs from prompt writing.
• I can explain why more context is not automatically better.
• I can classify sources by authority, currency, and relevance.
• I can build a source hierarchy with a conflict rule.
• I can construct a minimised, structured context package.
• I can distinguish authorised instructions from source content.
• I can identify when a task requires decomposition.
• I can decompose a complex task into logical stages.
• I can write stage-specific prompts using PCAFE-V.
• I can diagnose error propagation and place verification points.
• I can design structured inputs and outputs.
• I can define permission boundaries.
• I can recognise and respond to drift.
• I can identify and respond to prompt injection attempts.
• I can apply all five elements of SCOPE.
• I can identify revalidation triggers.

8. Looking Ahead: Unit 6

Congratulations—you have completed Unit 5! You now have a comprehensive understanding of advanced prompting, context engineering, and controlled workflow design.

In Unit 6: Human–AI Collaboration, you will build on this foundation to examine:

• Detailed classification of AI roles
• Human–AI complementarity
• Complete allocation of human and AI tasks
• Meaningful human oversight
• Automation bias
• Responsibility laundering
• Human-in-the-loop, human-on-the-loop, and human-out-of-the-loop arrangements
• Approval-authority design
• Reviewer capacity and independence
• The PARTNER framework
• Final decision and stop authority

You are now well-prepared to extend your Assignment 2 portfolio entries with context-engineered, multi-stage designs, and to reflect on your practice in Assignment 1.