COMP 3270 / Module 4 / Unit 12 / Lesson 12.1

Unit 12 · Lesson 12.1 · Week 8

Learning Goals, Assistance, and Substitution

Establish the foundational distinction between learning and task completion in the context of generative AI. Learn to identify learning goals, distinguish assistance from substitution, and design independent-attempt plans that preserve your own intellectual responsibility.

Estimated study time: 2.5–3.5 hours Primary reading: Chapter 12 Prerequisites: Units 1–11; TRACE & CLAIM

Table of Contents

Lesson Overview

This first lesson of Unit 12 establishes the foundational distinction between learning and task completion in the context of generative AI. You will examine how the same AI action can either support or substitute for learning, depending on the learning goal, the timing of assistance, and the learner's independent effort. You will learn to identify the intended learning outcome of an activity, determine which knowledge or skill must remain with the learner, and apply appropriate boundaries for AI use. By the end of this lesson, you will be equipped to design an independent-attempt plan that preserves your own intellectual responsibility while using AI as a constructive tool.

This lesson also reviews key verification concepts from Units 10 and 11 (TRACE and CLAIM), because learning from AI-generated material requires critical evaluation before you trust or internalise it. The skills developed here prepare you for Lesson 12.2 (productive struggle and scaffolding), Lesson 12.3 (retrieval and transfer), and Lesson 12.4 (the LEARN framework and personal workflows).

📘 Learning and academic-integrity reminder: This lesson is formative and ungraded. It is designed to help you build responsible AI-supported learning practices. Always follow your course, assignment, and institutional requirements. AI use that is appropriate for practice may be prohibited during formal assessments.

Lesson Learning Outcomes

By the end of this lesson, you will be able to:

  1. Define AI-supported learning and AI-substituted learning.
  2. Distinguish task completion from learning.
  3. Identify the intended learning outcome of an activity.
  4. Identify the knowledge, skill, judgment, or capability that should remain with the learner.
  5. Distinguish appropriate AI assistance from inappropriate substitution.
  6. Identify applicable assessment and academic-integrity boundaries.
  7. Select an appropriate AI role for a learning activity.
  8. Explain why answer-first AI use may weaken learning.
  9. Design an independent-attempt plan that preserves learner effort.
  10. Apply TRACE and CLAIM to evaluate AI-generated explanations.

Before You Begin

Retrieval Practice: Units 10 & 11

Before we begin new material, retrieve what you already know about verification and evidence. Attempt each question from memory before opening the answer.

Question 1. What is the primary purpose of the TRACE framework when evaluating an AI-generated explanation?

Show answer

Answer: TRACE is used to identify errors, omissions, misleading reasoning, and uncertainty in AI-generated content. It helps you classify the type of error (e.g., factual error, reasoning gap, hallucination) and determine what needs to be verified.

Explanation: TRACE stands for Types of error, Reasoning assessment, Assumption check, Claim verification, and Evidence evaluation. It is a diagnostic tool that precedes CLAIM.

Question 2. According to the CLAIM framework, what are the minimum criteria you should apply when verifying a claim made by an AI system?

Show answer

Answer: CLAIM requires that you: (1) identify the claim, (2) locate the source or origin of the claim, (3) assess the authority and reliability of the source, (4) check for corroborating or conflicting evidence, and (5) determine the strength of the claim given the available evidence.

Explanation: CLAIM helps you move from identifying a potential error (TRACE) to systematically verifying the claim using external sources. You should not accept any AI-generated factual claim without verification.

Question 3. Why is source verification necessary before you integrate an AI-generated explanation into your own learning?

Show answer

Answer: AI systems can produce plausible but incorrect, outdated, or misleading information. If you learn from an unverified explanation, you may internalise errors, develop misconceptions, or rely on fabricated sources. Verification ensures that the material you are learning is accurate and trustworthy.

Explanation: This is a core principle of epistemic responsibility: you are accountable for the knowledge you claim to have learned. Verification is not optional when using AI for learning.

Question 4. What is the difference between a hallucination and a reasoning error in an AI-generated response?

Show answer

Answer: A hallucination is content that is fabricated or invented without basis in the training data — the AI generates a plausible-sounding but false statement. A reasoning error occurs when the AI's logic or inference is flawed, even if the individual facts are correct.

Explanation: Hallucinations are often more difficult to detect because they are presented confidently and coherently. Reasoning errors may be easier to spot through careful analysis, but both require verification.

Question 5. What is one practical strategy you can use to verify a claim made by an AI system about a topic you are learning?

Show answer

Answer: One practical strategy is to consult a trusted, authoritative source — such as a textbook, peer-reviewed article, or official institutional document — and compare the AI's claim against that source. You can also cross-check with multiple sources and note any discrepancies.

Explanation: This is a core CLAIM practice. Other strategies include using the AI itself to locate sources (then verifying those sources), asking the AI to justify its claim with evidence, and using human subject-matter expertise where available.

Opening Scenario: Answer-First AI Use

Scenario

Maya is a third-year student in a health policy course. She has just received her assignment prompt: "Analyse the factors that influenced the adoption of electronic health records in Canada between 2010 and 2020. Use at least three scholarly sources and provide a critical assessment of implementation challenges."

Maya is pressed for time and feels uncertain about where to start. She opens an AI system and enters the following prompt: "Write a 1500-word essay analysing the factors that influenced the adoption of electronic health records in Canada between 2010 and 2020, with at least three scholarly sources." The AI produces a well-structured essay with citations. Maya reads it, feels she understands the topic, and submits the essay as her own work. She receives a passing grade but later realises that she cannot recall the main arguments, explain the factors she "analysed," or identify the sources used in the essay.

Questions for reflection:

  1. What was the intended learning outcome of Maya's assignment?
  2. Which work should have remained with Maya as the learner?
  3. What assistance might have been appropriate, and at what point?
  4. What policy or disclosure questions arise from Maya's approach?
  5. How could Maya's interaction with AI be redesigned to support — rather than replace — her learning?
Show guided analysis

Guided analysis:

  • Intended learning outcome: The assignment was designed to develop Maya's ability to analyse a complex policy issue, synthesise scholarly sources, and articulate a critical argument in her own words. The learning outcome was not merely to produce a formatted essay but to develop analytical, research, and writing skills.
  • Work that should remain with Maya: The analysis, synthesis, critical evaluation, source selection, and argument construction are the core learning activities. The AI should not perform these on Maya's behalf.
  • Appropriate assistance: Maya could have used AI to generate a list of search terms, suggest a reading list (then verify), explain a difficult concept from the readings, provide feedback on a draft outline, or ask clarification questions. All of these would preserve her intellectual responsibility.
  • Policy and disclosure: Maya likely violated her institution's academic-integrity policy by submitting AI-generated work as her own. Even if the AI output was not detected, she did not disclose her use of the tool, and she presented the AI's analysis as her own critical work.
  • Redesign: Maya should begin by reading and annotating key sources, then forming her own initial arguments. She could then use AI to check her understanding, ask for counterarguments, or get feedback on her draft. She should document her AI use and disclose it according to course policy.

Key insight: Maya's approach is an example of AI-substituted learning because the AI performed the cognitive work she was expected to develop. Her recognition of the AI's output created an illusion of understanding that did not translate into independent recall or application.

Core Content

1. Learning Goals: The Foundation of Responsible AI Use

A learning goal describes the knowledge, skill, understanding, judgment, or capability the learner is expected to develop. Every responsible use of AI for learning begins with a clear learning goal. Without a learning goal, you cannot determine whether AI assistance is supportive or substitutive.

Learning goals may involve:

🔑 Key principle: AI use should be evaluated against the learning goal. If the AI performs the cognitive work that the learning goal requires the learner to develop, the use is likely substitutive rather than supportive.

Example: If the learning goal is to explain the concept of cognitive offloading in your own words, then asking the AI to generate an explanation for you to copy would be substitutive. Asking the AI to generate a question that prompts you to explain the concept would be supportive.

Common misconception: "If the AI produces a correct answer, I have learned something." — This is not necessarily true. Recognition of a correct answer is not the same as understanding, recall, or independent application. Learning requires active processing by the learner, not passive receipt of information.

2. Task Completion versus Learning

Task completion means producing a required output or answer. It is the visible product of an activity — the essay, the solution set, the code, the report. Task completion can occur without any learning, and learning can occur without any visible task completion.

A completed task may not demonstrate:

⚠️ Critical warning: A correct-looking answer is not evidence that learning occurred. Many students have completed assignments using AI and later discovered they could not explain the content or solve a related problem independently. The task was completed, but the learning did not happen.

Example: A student uses AI to solve a set of calculus problems. The student submits the solutions and receives full credit. However, when the student encounters a similar problem on an exam, they cannot solve it without the AI. The task was completed, but the learning did not occur.

Why this matters: In courses, assessments are designed to measure learning, not just task completion. If you use AI to complete the task without engaging in the learning process, you may pass the assignment but fail the exam — and more importantly, you will not have developed the capability you are expected to carry forward into your professional life.

3. AI-Supported Learning versus AI-Substituted Learning

The distinction between AI-supported learning and AI-substituted learning is not about the AI tool itself, but about how it is used in relation to the learning goal.

AI-Supported Learning

AI-supported learning occurs when AI assistance helps the learner engage in and strengthen the intended learning activity. The learner remains responsible for understanding, verification, reconstruction, and application.

Examples of AI-supported learning include:

AI-Substituted Learning

AI-substituted learning occurs when AI performs the cognitive, creative, communicative, analytical, or reflective work that the learner is expected to develop or demonstrate.

Examples of AI-substituted learning include:

📌 Critical distinction: The same AI action may be supportive in one context and substitutive in another. For example, asking AI to generate a list of search terms for a literature review may be supportive; asking AI to write the entire literature review is substitutive. The difference is determined by the learning goal, the timing of assistance, and the amount of learner effort.

4. Assistance versus Substitution: A Decision Framework

To determine whether a particular AI use is assistance or substitution, examine these factors:

Assistance vs. Substitution: A comparison
Dimension Assistance (Supportive) Substitution (Replace)
Learner effort Learner engages actively; performs core reasoning. AI performs core reasoning; learner is passive.
Timing After an independent attempt. Before an independent attempt (answer-first).
AI output Hints, questions, feedback, partial examples, alternative explanations. Complete answers, assessed writing, full analyses, personal reflections.
Learner's role Verifier, reconstructor, transferer, reflecter. Consumer, submitter, editor (at best).
Evidence of learning Learner can explain, reconstruct, apply, and transfer. Learner may recognise the output but cannot reproduce it.

5. The Effect of Answer-First AI Use

Using AI before making an independent attempt — often called "answer-first" AI use — has specific effects on learning:

⚠️ Research insight: Learning science consistently shows that retrieval practice — attempting to recall information from memory — is one of the most powerful learning strategies. Answer-first AI use bypasses retrieval and weakens the learning process. The learner may complete the task but will not have developed the underlying capability.

Example: Two students are learning about the concept of "transfer" in learning science. Student A asks the AI: "Explain transfer in learning science." The AI provides a clear definition and examples. Student A reads it and moves on. Student B attempts to define transfer from memory, then uses the AI to check and refine their understanding. Student B has engaged in retrieval practice and is more likely to retain and apply the concept.

6. AI Roles for Learning

AI can assume different roles in a learning interaction. The appropriateness of each role depends on the learning goal and the boundary conditions.

Tool:
AI acts as a resource that the learner uses to perform a task (e.g., generating a list of search terms, formatting a citation). The learner retains full control and decision-making.
Assistant:
AI supports the learner by providing hints, questions, or partial guidance. The learner remains the primary agent and does the core cognitive work.
Critic:
AI evaluates the learner's work and provides feedback. The learner must assess the feedback and decide whether and how to act on it.
Simulator:
AI creates a practice environment (e.g., a simulated conversation, a practice problem set) in which the learner applies skills independently.
Recommender:
AI suggests resources, strategies, or next steps. The learner evaluates the recommendations and makes the final decision.
Agent:
AI acts autonomously to complete a task on the learner's behalf (e.g., writing an essay, generating a solution). This is generally inappropriate for learning unless the learner is not expected to develop that capability.
💡 Guiding principle: For a learning activity, the most appropriate AI roles are those that empower the learner to engage actively with the material. Roles that replace the learner's thinking are generally substitutive and should be avoided or used only in practice contexts where no assessment is involved.

7. Assessment Boundaries and Academic Integrity

Every course, assignment, and institution has specific rules about AI use. It is your responsibility to understand and follow these boundaries. A use that is appropriate for practice may be prohibited during an assessment.

Key boundary questions to ask:

📌 Important: When in doubt, ask your instructor. Do not assume that AI use is permitted just because it is possible. Many instructors explicitly prohibit AI use for certain types of assessments, and using AI in violation of these rules is a form of academic misconduct.

Example: In a writing course, the instructor may allow AI for brainstorming and editing but prohibit AI for generating the core arguments. In a mathematics course, the instructor may prohibit AI entirely for problem sets. Always check the specific instructions.

8. Designing an Independent-Attempt Plan

Before you use AI for a learning activity, you should have a clear plan for what you will attempt independently. An independent-attempt plan specifies:

An independent-attempt plan preserves your responsibility as a learner and ensures that the AI supports — rather than replaces — your learning.

✅ Key practice: Always begin a learning activity by attempting it yourself. This may be uncomfortable if you are unsure, but that discomfort is part of the learning process. The effort you invest before seeking help strengthens your understanding and makes the AI's assistance more effective when you do use it.

Learning-Process Walkthrough: From Learning Goal to Independent Attempt to Bounded Assistance

This walkthrough illustrates a responsible learning process using AI. Follow the steps to see how a learner moves from identifying the learning goal to making an independent attempt to using bounded AI assistance.

Context: A student in a business ethics course is learning about the concept of "stakeholder theory." The learning goal is to explain stakeholder theory and apply it to a real-world business scenario.
  1. Identify the learning goal: The student reads the course material and identifies the specific outcome: "By the end of this activity, I should be able to define stakeholder theory, identify its key principles, and apply it to a business case study."
  2. Identify what must remain with the learner: The student recognises that the explanation and application must come from them. The AI can help with clarification and examples, but the student must do the core reasoning.
  3. Make an independent attempt: The student writes a short definition of stakeholder theory in their own words, based on the course reading. They also sketch a rough application to a familiar company (e.g., a local business they know).
  4. Diagnose the difficulty: The student realises they are unsure about how to distinguish stakeholder theory from shareholder theory. They also struggle to identify which stakeholders are most relevant in their example.
  5. Request bounded assistance: The student asks the AI: "What is the key difference between stakeholder theory and shareholder theory? Please provide a brief comparison without giving a full case analysis." This request is specific, bounded, and preserves the student's responsibility for the application.
  6. Verify the AI response: The student uses TRACE and CLAIM to check the AI's explanation against the course textbook and a credible business ethics source. They find that the AI's comparison is accurate but lacks nuance about Freeman's original work. The student notes this and supplements the AI's explanation with their own reading.
  7. Reconstruct and apply: The student revises their definition and application, incorporating the verified information from the AI and the course material. They now feel confident that they can explain stakeholder theory and apply it to a new case.
  8. Document and reflect: The student records the AI prompt, the response, and their verification process. They reflect on what they learned and what they still need to explore.
✅ Outcome: The student engaged with the material, used the AI to resolve a specific difficulty, verified the AI's contribution, and produced their own explanation and application. The learning goal was achieved, and the student retained the capability.
Show solution design: Key decisions
  • Learning goal: Define and apply stakeholder theory.
  • Independent attempt: Write a definition and apply it to a known company.
  • AI role: Assistant and critic (clarifying the distinction between theories).
  • Bounded assistance: A specific, narrow question that preserves the learner's core work.
  • Verification: TRACE and CLAIM used to check the AI's explanation.
  • Reconstruction: The learner revises their work based on verified information.
  • Documentation: AI use is recorded and reflected upon.

This design ensures that the AI supports the learning without substituting for it.

Worked Example: Applying the Distinction in a Realistic Scenario

This worked example demonstrates the full process of distinguishing learning from task completion, using a synthetic but realistic scenario.

Scenario

A student in a public administration course is learning about "policy feedback theory." The assignment requires the student to explain the concept of policy feedback and illustrate it with an example from Canadian social policy. The student decides to use AI to support their learning.

Worked Example: Learning vs. Task Completion
Step Student Action AI Role Learning or Substitution?
1. Learning goal Student identifies the goal: "Explain policy feedback and provide a Canadian example." — Learning (goal setting)
2. Independent attempt Student writes a preliminary definition based on course reading and recalls an example from class (the Canada Child Benefit). — Learning (independent recall)
3. Diagnostic question Student asks: "Does the Canada Child Benefit illustrate policy feedback? I think it does, but I'm not sure how to explain the mechanism." Assistant: provides a clarifying explanation of policy feedback mechanisms and asks the student to identify the mechanism in the Canada Child Benefit. Learning (supported)
4. Verification Student checks the AI's explanation against the course textbook and a peer-reviewed article. The AI's explanation is accurate but omits one key mechanism (positive feedback). — Learning (verification)
5. Reconstruction Student revises their explanation, incorporating the verified information and adding the missing mechanism. — Learning (reconstruction)
6. Application Student applies the concept to a new example (the Canada Pension Plan) and writes a concise explanation. — Learning (transfer)
7. Documentation Student records the AI prompt, the response, and their verification notes. — Learning (reflection)
Show analysis

Analysis: In this worked example, the student used AI as an assistant to clarify a specific difficulty (the mechanism of policy feedback). The student retained responsibility for the definition, the example selection, the verification, the reconstruction, and the application. The AI did not generate the final explanation or the example; the student did. This is an example of AI-supported learning.

What would substitution look like here? If the student had asked the AI to "write a complete explanation of policy feedback with a Canadian example" and submitted the AI's output, that would have been AI-substituted learning — the AI performed the explanation, analysis, and example selection, and the student merely submitted the output.

Guided Learning Activities

Activity 1: Learning Goal or Output Goal?

Purpose: Distinguish between goals that describe learning and goals that describe outputs. · Time: 15 min · Outcome: Identify the intended learning outcome of an activity.

Instructions: Read each statement below and classify it as:

  • (L) Learning goal
  • (O) Output goal
  • (A) Assessment requirement
  • (S) Learning strategy
  • (I) Insufficient information
  1. "Write a 2000-word essay on the economic impact of the pandemic."
  2. "Analyse the economic impact of the pandemic using at least three theoretical frameworks."
  3. "Submit a completed problem set with all 10 questions answered."
  4. "Practice explaining the concept to a peer to check your understanding."
  5. "Develop the ability to evaluate competing economic models."
  6. "Use AI to generate a list of possible topics for your research paper."
Show classification and explanation
  1. O (Output goal): Specifies the format and length of the output, not what the learner will develop.
  2. L (Learning goal): Describes the skill to be developed (analysis) and the content (economic impact, theoretical frameworks).
  3. A (Assessment requirement): Specifies what must be submitted, not what the learner will learn.
  4. S (Learning strategy): Describes a strategy for learning (self-explanation), not a goal.
  5. L (Learning goal): Describes the development of a capability (evaluating competing models).
  6. I (Insufficient information): The role of AI is mentioned, but the learning goal of the activity is not specified. It could be supportive or substitutive depending on the context.

Key insight: Learning goals describe what the learner will be able to do, while output goals describe what will be produced. Assessment requirements specify what must be submitted, and learning strategies describe how to achieve a goal.

Activity 2: Assistance or Substitution?

Purpose: Classify AI uses as supportive assistance, substitution, or policy-dependent. · Time: 20 min · Outcome: Distinguish appropriate AI assistance from inappropriate substitution.

Instructions: For each scenario, classify the AI use as:

  • (SA) Supportive assistance
  • (PO) Possible over-assistance
  • (IS) Inappropriate substitution
  • (PP) Permitted only in practice (not for assessment)
  • (PD) Policy dependent
  • (II) Insufficient information
  1. A student uses AI to generate a list of search terms before starting a literature review.
  2. A student asks AI to rewrite a paragraph they wrote to improve clarity and style.
  3. A student asks AI to generate the entire introduction section of their research paper.
  4. A student uses AI to create practice problems for a concept they are learning.
  5. A student uses AI to check their answer to a mathematics problem after solving it independently.
  6. A student uses AI to complete a quiz that is part of their final grade.
Show classification and explanation
  1. SA: Generating search terms helps the student begin the research process without replacing their analysis of the literature.
  2. SA / PD: Editing assistance can be supportive if the student wrote the original draft. However, some courses may prohibit AI editing. Check the policy.
  3. IS: The AI is generating the core intellectual content of the paper. This is substitution unless the learning goal is to evaluate AI-generated text (which is unlikely).
  4. SA / PP: Creating practice problems is a supportive use for learning, but if those problems are used for graded assessment, the context changes.
  5. SA: The student solved the problem independently and uses AI to check their work. This is supportive because the learning already occurred.
  6. IS / PD: Completing a graded quiz using AI is almost always prohibited. Even if not explicitly prohibited, it is a substitution because the assessment is designed to measure the student's independent capability.

Key insight: The same AI action can be supportive in one context and substitutive in another. The learning goal, timing, and assessment rules all matter.

Activity 3: AI Role for Learning

Purpose: Select an appropriate AI role for a given learning goal. · Time: 15 min · Outcome: Select an appropriate AI role for a learning activity.

Instructions: For each learning goal, identify which AI role(s) would be appropriate. Roles: Tool, Assistant, Critic, Simulator, Recommender, Agent.

  1. Learning goal: "Generate and evaluate alternative approaches to solving a complex problem."
  2. Learning goal: "Recall and explain the key principles of the Canadian Charter of Rights and Freedoms."
  3. Learning goal: "Write a persuasive argument on a policy issue, using evidence and reasoning."
  4. Learning goal: "Practice conducting a mock job interview."
Show recommendations and explanation
  1. Assistant, Critic, Recommender: The AI can assist by generating alternative approaches, critique the learner's evaluation, and recommend resources. The learner should do the evaluation and decision-making. Agent would be inappropriate.
  2. Assistant, Simulator: The AI can ask questions to prompt recall (simulator) and provide feedback on the learner's explanation (assistant). The learner must do the recall and explanation. Agent would be substitution.
  3. Critic, Assistant: The AI can provide feedback on the learner's draft (critic) and suggest alternative evidence (assistant). The learner must write the argument. Agent would be substitution.
  4. Simulator: The AI can role-play as an interviewer (simulator). The learner must practice their responses. Agent would be inappropriate.

Key insight: In all cases, the Agent role — where the AI performs the task autonomously — is inappropriate for learning because it replaces the learner's development of the capability.

Activity 4: Independent-Attempt Plan

Purpose: Design an independent-attempt plan for a learning activity. · Time: 20 min · Outcome: Design an independent-attempt plan that preserves learner effort.

Instructions: For the following learning activity, write a brief independent-attempt plan. The activity is: You are learning about the concept of "confirmation bias" in a psychology course. You need to be able to explain the concept and identify it in a real-world situation.

Your plan should address: what you will attempt independently; how long or how deeply you will attempt; what evidence of effort you will retain; when you will seek a hint; when you will consult a human; what AI assistance remains outside scope.

Show solution design

Sample independent-attempt plan:

  • What to attempt independently: Read the course material on confirmation bias. Write a one-paragraph definition in my own words. Identify three real-world examples from my own experience.
  • How long/deep: Spend 30 minutes reading and annotating. Spend 15 minutes writing the definition. Spend 10 minutes generating examples. Total: 55 minutes of independent work.
  • Evidence of effort: Annotated reading notes, draft definition, list of examples.
  • When to seek a hint: If I cannot identify a clear example, or if I am unsure whether my definition captures the key elements, I will ask the AI: "What are the key elements of confirmation bias that I should include in my definition?"
  • When to consult a human: If my definition still feels wrong after using the AI hint, or if I am confused about the difference between confirmation bias and other cognitive biases, I will ask my tutor or the course discussion forum.
  • AI assistance outside scope: I will not ask the AI to write my definition for me. I will not ask the AI to generate the examples for me. I will only use it for clarification and feedback.

Exercises

Complete these exercises to reinforce your understanding. Attempt each one before reviewing the answer.

Exercise 1: Distinguish Learning Goals from Output Goals

Outcome: Distinguish learning goals from output goals. · Time: 15 min.

Task: For each statement, classify it as a learning goal (LG), an output goal (OG), an assessment requirement (AR), a learning strategy (LS), or insufficient information (II). Justify your classification.

  1. "I will demonstrate my ability to design a responsible AI use policy for a small organisation."
  2. "I will create a 10-slide presentation on responsible AI use."
  3. "I will practice explaining the AI Paradox to a non-technical audience."
  4. "I will write a 500-word summary of Chapter 12."
  5. "I will develop confidence in my ability to evaluate AI-generated claims."
  6. "I will use AI to generate a draft of my assignment."
Show answers and explanations
  1. LG: Describes the capability to be developed (designing a policy).
  2. OG: Specifies the format and length of the output, not the capability.
  3. LS: Describes a practice activity, not a goal.
  4. OG: Specifies the output (summary) and length.
  5. LG: Describes the development of confidence and evaluation ability.
  6. II: The statement describes an AI action, but the learning goal of the activity is not specified. It could be supportive or substitutive depending on context.
Exercise 2: Identify the Capability Being Assessed

Outcome: Identify the capability being assessed. · Time: 15 min.

Task: For each assessment task, identify the capability being assessed (e.g., recall, analysis, synthesis, evaluation, application, creation, communication, judgment, reflection).

  1. "Write an essay analysing the causes of the First World War."
  2. "Solve the following 10 mathematics problems without using a calculator."
  3. "Participate in a simulated client consultation and provide professional advice."
  4. "Create a portfolio of your original artwork with a reflective commentary."
  5. "Answer 20 multiple-choice questions about the Canadian legal system."
Show answers and explanations
  1. Analysis, synthesis, communication.
  2. Application, procedural skill.
  3. Application, professional judgment, communication.
  4. Creation, reflection, communication.
  5. Recall, recognition.

Key insight: Identifying the capability being assessed helps you determine whether AI assistance is appropriate. If the capability is one you need to develop, the AI should not perform it for you.

Exercise 3: Classify Assistance and Substitution

Outcome: Classify AI uses as assistance or substitution. · Time: 20 min.

Task: For each scenario, classify as (SA) supportive assistance, (PO) possible over-assistance, (IS) inappropriate substitution, or (PD) policy dependent.

  1. A student uses AI to generate a bibliography for their research paper, then verifies each source.
  2. A student uses AI to write a personal reflection on their learning progress.
  3. A student uses AI to check the grammar and spelling of their final submission.
  4. A student uses AI to generate the main arguments for their thesis, then revises them slightly.
  5. A student uses AI to create flashcards for a vocabulary quiz.
Show answers and explanations
  1. SA / PD: Generating a bibliography can save time, but some courses may prohibit AI for any part of the research process. The verification step is crucial.
  2. IS: A personal reflection is meant to come from the learner's own experience and thinking.
  3. SA / PD: Checking grammar and spelling is a mechanical task that does not replace the core intellectual work.
  4. IS: Generating the main arguments replaces the core analytical work of the thesis.
  5. SA / PP: Creating flashcards is supportive for learning, but if used for a graded quiz, the context changes.
Exercise 4: Select an Appropriate AI Role

Outcome: Select an appropriate AI role. · Time: 15 min.

Task: For each learning goal, identify which AI role(s) are appropriate and which are inappropriate.

  1. Learning goal: "Develop the ability to write persuasive arguments in a professional context."
  2. Learning goal: "Learn to interpret statistical output and draw conclusions."
  3. Learning goal: "Prepare for a high-stakes negotiation simulation."
Show answers and explanations
  1. Appropriate: Critic, Assistant, Recommender. Inappropriate: Agent.
  2. Appropriate: Assistant, Simulator, Critic. Inappropriate: Agent.
  3. Appropriate: Simulator, Assistant, Critic. Inappropriate: Agent.
Exercise 5: Define an Independent Attempt

Outcome: Define an independent attempt. · Time: 20 min.

Task: You are learning about "confirmation bias" in a psychology course. Design an independent-attempt plan.

Show solution design
  • Independent attempt: Read the textbook section, write a definition in my own words, identify three examples.
  • Time: 40 minutes.
  • Evidence: Annotated textbook, written definition, list of examples.
  • Hint trigger: If I cannot identify a clear example after 10 minutes, or if I am unsure about the key elements of the definition.
  • Outside scope: I will not ask AI to write my definition or generate examples for me.
Exercise 6: Apply an Assessment and Disclosure Boundary

Outcome: Apply assessment and disclosure boundaries. · Time: 15 min.

Task: A course policy says: "AI may be used for brainstorming and editing, but the core analysis and writing must be your own. All AI use must be disclosed." For each use, determine whether it is permitted and what disclosure is required.

  1. Ask AI to generate a list of potential topics.
  2. Ask AI to write the first draft of the introduction.
  3. Use AI to check grammar and spelling.
  4. Ask AI to suggest counterarguments to your thesis.
  5. Ask AI to generate the full analysis section.
Show answers and explanations
  1. Permitted: Brainstorming is allowed. Disclosure is required.
  2. Not permitted: This is substitution, not brainstorming or editing.
  3. Permitted: Editing is allowed. Disclosure is required.
  4. Permitted: Brainstorming counterarguments is allowed. Disclosure is required.
  5. Not permitted: This is substitution.
Exercise 7: Identify Answer-First AI Use

Outcome: Explain why answer-first AI use may weaken learning. · Time: 15 min.

Scenario: A student in a philosophy course is learning about "utilitarianism." Before reading the course material, the student asks the AI: "What is utilitarianism? Provide a comprehensive explanation with examples." The AI provides a detailed explanation. The student reads it and moves on to the next topic.

Questions:

  1. Is this AI-supported or AI-substituted learning? Why?
  2. What effect might this approach have on the student's ability to recall and apply the concept?
  3. How could the student redesign their approach to use AI supportively?
Show answers and explanations
  1. AI-substituted learning. The AI performed the explanation before the student made any independent attempt.
  2. Effect: The student may recognise the concept but may not be able to recall or apply it independently.
  3. Redesign: The student could first read the course material and attempt to define utilitarianism in their own words, then use the AI to check their understanding.

Homework Problems

Formative · Ungraded · Recommended learning practice; may serve as evidence for Assignment 1.

📝 Note: Each problem includes hints and an answer/solution design. Use the hints only after you have made a genuine attempt.
Problem 1: Learning Goal Analysis

Task: A course learning outcome states: "Students will be able to design and justify a research methodology appropriate for a social science question." Is this a learning goal or an output goal? Justify your answer.

Show answer and explanation

Answer: This is a learning goal. It describes the capability the learner will develop (designing and justifying a research methodology). It does not specify the format of the output.

Problem 2: Task Completion vs. Learning

Task: A student uses AI to generate a complete solution to a complex statistics problem. The student submits the solution and receives full credit. Has learning occurred? Why or why not?

Show answer and explanation

Answer: Learning may not have occurred. The task was completed, but the student did not engage in the cognitive work of solving the problem. Without independent recall and application, the student likely did not develop the capability.

Problem 3: Assistance vs. Substitution

Task: A student is learning to write JavaScript code. The student uses AI to generate code for a function, then studies the code, modifies it, and writes a similar function from memory. Classify this use.

Show answer and explanation

Answer: This is supportive assistance. The student used the AI to generate an example, then actively studied, modified, and reconstructed the code from memory.

Problem 4: AI Role Selection

Task: A student is learning to present complex data to a non-technical audience. What AI role(s) would be appropriate? Which would be inappropriate?

Show answer and explanation

Appropriate: Assistant, Critic, Recommender, Simulator.

Inappropriate: Agent.

Problem 5: Independent-Attempt Plan Design

Task: You are learning to apply the principles of "design thinking" to a real-world problem. Write a brief independent-attempt plan.

Show solution design
  • Independent attempt: Identify a real-world problem and apply the five stages of design thinking.
  • Time: 90 minutes.
  • Evidence: Notes on each stage, sketches of ideas, a rough prototype.
  • Hint trigger: If I get stuck at a specific stage.
  • Outside scope: I will not ask AI to design the solution for me.
Problem 6: Policy Interpretation

Task: A course policy states: "AI may be used for research assistance but not for completing any graded assessment." A student uses AI to generate a summary of a scholarly article, then reads the summary and uses it to inform their own analysis. Is this use permitted?

Show answer and explanation

Answer: Likely permitted, provided the student does not submit the summary as their own work and uses it only for research assistance. The student should verify the summary using the original source (CLAIM).

Problem 7: Disclosure Requirement

Task: A student uses AI to help them brainstorm ideas for an essay. The student writes the essay independently but uses the AI-generated ideas as a starting point. What should the student disclose?

Show answer and explanation

Answer: The student should disclose that they used AI for brainstorming. A sample disclosure: "I used AI to generate a list of potential topics and ideas for this essay. All analysis, argumentation, and writing are my own."

Problem 8: No-AI Boundary

Task: An exam is proctored and students are required to complete it without any external assistance. A student uses AI to check their understanding the night before the exam but does not use it during the exam. Is this permitted?

Show answer and explanation

Answer: This is permitted, as long as the student does not use AI during the exam. Using AI for preparation is a form of learning support.

Problem 9: Transfer Problem

Task: Apply the concepts of AI-supported and AI-substituted learning to a professional workplace setting. Describe one example of each.

Show solution design

AI-supported: A junior analyst uses AI to generate a list of potential data sources for a market analysis report, then independently evaluates each source and conducts the analysis.

AI-substituted: A marketing professional asks AI to generate a complete social media strategy, then presents it as their own work.

Problem 10: Comprehensive Analysis

Task: A student in a nursing course is learning about patient communication techniques. The student uses AI to generate a script for a patient consultation, then studies the script, practices it, and records themselves performing a consultation. Analyse this use.

Show answer and explanation
  • Learning vs. task completion: This is learning. The student used the script as a resource but engaged in active practice and reconstruction.
  • Assistance vs. substitution: This is assistance. The AI generated a script (a tool), but the student did the learning work.
  • AI role: Tool and Simulator.
  • Disclosure: The student should disclose that they used AI to generate a practice script.

Mini-Project: AI-Supported Learning Suitability Plan

Purpose

Apply the concepts from this lesson to a real or simulated learning activity by developing a comprehensive suitability plan. Estimated effort: 90–120 minutes.

Task

Select one bounded learning activity from your own studies, or use one of the following synthetic contexts:

Develop a suitability plan that addresses the following 15 elements:

  1. Subject or discipline
  2. Learning goal
  3. Knowledge or skill to be developed
  4. Expected learner evidence
  5. Independent attempt
  6. Proposed AI role
  7. Permitted AI assistance
  8. Prohibited or unsuitable assistance
  9. Assessment boundary
  10. Verification requirement
  11. Human assistance option
  12. Privacy boundary
  13. Documentation requirement
  14. Stop condition
  15. Preliminary recommendation
Show staged hints

Hint 1: Begin by defining the learning goal clearly. What specific capability do you want to develop?

Hint 2: Design your independent attempt to be specific and challenging. What will you do first?

Hint 3: Be specific about what AI assistance is permitted and what is prohibited.

Hint 4: Plan how you will verify AI-generated content (TRACE and CLAIM) and how you will document and disclose your AI use.

Show solution architecture and partial example

Partial example (carbon footprint analysis):

  1. Subject: Environmental studies / sustainability.
  2. Learning goal: Calculate the carbon footprint of a small organisation and identify reduction strategies.
  3. Knowledge or skill: Applying emission factors, interpreting data, making recommendations.
  4. Learner evidence: Written analysis with calculations and recommendations.
  5. Independent attempt: Gather data, calculate emissions, propose three reduction strategies.
  6. AI role: Assistant (checking calculations, suggesting strategies).
  7. Permitted: Checking calculations, generating potential strategies for evaluation.
  8. Prohibited: AI generating the analysis or recommendations.
  9. Assessment boundary: Practice activity.
  10. Verification: Check emission factors against official sources.
  11. Human assistance: Consult the tutor if calculation methods are unclear.
  12. Privacy: Do not enter the organisation's actual financial data.
  13. Documentation: Record all AI prompts and outputs.
  14. Stop condition: Stop if AI begins generating the analysis.
  15. Recommendation: Proceed with verification.
✅ Self-review checklist:
  • Did I clearly define the learning goal?
  • Did I specify what I will do independently?
  • Did I distinguish between permitted and prohibited AI assistance?
  • Did I plan for verification using TRACE and CLAIM?
  • Did I consider privacy, disclosure, and stop conditions?
  • Is my plan realistic and actionable?

Self-Check Quiz

Test your understanding with these questions. Attempt each from memory before revealing the answer.

1. What is the primary distinction between AI-supported learning and AI-substituted learning?

Show answer and explanation

Answer: AI-supported learning occurs when AI helps the learner engage in and strengthen the intended learning activity, while AI-substituted learning occurs when AI performs the cognitive work that the learner is expected to develop.

Explanation: The distinction is not about the tool but about the learner's role and effort.

2. A student uses AI to generate a list of potential research topics before writing a proposal. This is most likely an example of:

Select one:
Show answer and explanation

Answer: B) AI-supported learning (or D, depending on context).

Explanation: Generating a list of topics is a brainstorming activity that supports the student's own work. Context matters.

3. Which of the following is not a sign that a completed task does not demonstrate learning?

Select one:
Show answer and explanation

Answer: B. If the learner can solve a similar problem independently, that is evidence that learning has occurred.

4. Which AI role is generally inappropriate for a learning activity because it replaces the learner's cognitive work?

Select one:
Show answer and explanation

Answer: C) Agent. The Agent role performs tasks autonomously on behalf of the learner, replacing the learner's cognitive work.

5. What is the primary risk of answer-first AI use in learning?

Select one:
Show answer and explanation

Answer: B) It reduces productive struggle and weakens retrieval practice.

6. A student uses AI to check their answer to a mathematics problem after solving it independently. This is an example of:

Select one:
Show answer and explanation

Answer: B) AI-supported learning. The student solved the problem independently first.

7. Which factor is least relevant to determining whether a particular AI use is assistance or substitution?

Select one:
Show answer and explanation

Answer: C) The length of the AI output. Length is not a reliable indicator.

8. When should a student stop using AI for a learning activity? (Select all that apply)

Select all that apply:
Show answer and explanation

Answer: A, B, and C. D (perfect formatting) is not a stop condition.

9. A student asks the AI: "What is the key difference between stakeholder theory and shareholder theory?" The student then uses the AI's response to write their own explanation. This is an example of:

Select one:
Show answer and explanation

Answer: B) AI-supported learning (provided the student writes their own explanation and verifies the AI's information).

Assignment Connection

Assignment 5: Verification and Fact-Checking Assignment (weight: 10%)

The verification skills you developed in Assignment 5 — using TRACE to identify errors and CLAIM to verify claims — are directly applicable to AI-supported learning. Apply these skills to every AI-generated explanation, hint, or piece of feedback you use for learning.

Assignment 1: AI Learning and Reflection Journal (weight: 15%)

This lesson provides evidence you can preserve for Assignment 1, including: your independent-attempt plans, classifications of assistance vs. substitution, reflections on answer-first AI use, AI role selections, and documentation of AI prompts and outputs.

Consider adding: your independent-attempt plan (Activity 4), your AI-Supported Learning Suitability Plan (Mini-Project), and your reflections on the opening scenario.

Assignment 1 Journal Connection

Consider the following prompts for your Assignment 1 journal entry. Reflect honestly on your own practices and experiences.

📓 Journal prompts
  1. Task completion vs. learning: Think of a time when you completed a task but did not learn the underlying material. How did that feel? What did you do differently (or wish you had done differently)?
  2. Assistance vs. substitution: Reflect on a time when you used AI for learning. Was your use more like assistance or substitution? What would you do differently now?
  3. Answer-first AI use: Have you ever used AI before making an independent attempt? What effect did that have on your learning? What would you change?
  4. Assessment boundaries: How do you determine whether AI use is permitted in a particular assignment? What strategies can you use to stay within boundaries?

Note: Do not submit these reflections as part of this lesson. They are prompts for your Assignment 1 journal.

Show reflection-planning checklist
  • Did you identify a specific instance of task completion without learning?
  • Did you analyse whether your AI use was assistance or substitution?
  • Did you consider the effect of answer-first AI use?
  • Did you reflect on how you determine assessment boundaries?
  • Did you identify any changes you will make to your practice?

Key Takeaways

Remember these points
  • Learning goals describe what you are expected to develop; output goals describe what you will produce.
  • Task completion is not evidence of learning. A correct answer does not necessarily mean you understand, recall, or can apply the material.
  • AI-supported learning occurs when AI helps you engage with and strengthen the intended learning activity. AI-substituted learning occurs when AI performs the cognitive work you are expected to develop.
  • The same AI action can be supportive in one context and substitutive in another.
  • Answer-first AI use reduces productive struggle, weakens retrieval, and creates an illusion of understanding. Always attempt the task independently first.
  • AI roles such as Assistant, Critic, Simulator, and Recommender can support learning. The Agent role is generally inappropriate.
  • You are responsible for understanding and following assessment boundaries.
  • An independent-attempt plan preserves your learning responsibility.
  • Always verify AI-generated content using TRACE and CLAIM.
  • Documentation and disclosure are essential components of responsible AI use.

Key Terms

AI-supported learning
AI assistance that helps the learner engage in and strengthen the intended learning activity, with the learner remaining responsible for understanding, verification, reconstruction, and application.
AI-substituted learning
AI assistance that performs the cognitive, creative, communicative, analytical, or reflective work that the learner is expected to develop or demonstrate.
Learning goal
A description of the knowledge, skill, understanding, judgment, or capability the learner is expected to develop.
Task completion
Producing a required output or answer, which may not demonstrate understanding, recall, independent skill, transfer, or judgment.
Assistance
AI use that supports the learner's own cognitive work without performing it on their behalf.
Substitution
AI use that performs the cognitive work the learner is expected to develop or demonstrate.
Answer-first AI use
Using AI to obtain an answer before making an independent attempt, which reduces productive struggle and weakens retrieval practice.
Independent-attempt plan
A structured plan specifying what the learner will do without AI, how long they will try, what evidence to retain, and when to seek assistance.
AI role
The function AI performs in a learning interaction: Tool, Assistant, Critic, Simulator, Recommender, or Agent.
Agent (AI role)
An AI role that acts autonomously to complete a task on the learner's behalf; generally inappropriate for learning.
Assessment boundary
The rules and constraints governing what AI use is permitted for a given assessment, including disclosure requirements.
Verification
The process of checking AI-generated content for accuracy, completeness, and reliability using frameworks such as TRACE and CLAIM.
Documentation
The practice of recording AI prompts, outputs, verification, corrections, and reflections for transparency and future reference.

Looking Ahead

In Lesson 12.2: Productive Struggle, Scaffolding, and Feedback, you will build on the foundation established here. You will learn how to design and use graduated hints, scaffolding, and formative feedback to support learning without replacing it. You will also learn to evaluate AI-generated feedback using TRACE and CLAIM, and to design scaffold fading that gradually reduces support as your competence increases.

After completing all four lessons in Unit 12, you will be prepared for Unit 13: Research and Scholarly Practice.

Continue to Lesson 12.2 Return to Unit 12 Home

Sources and Further Reading

[INSERT VERIFIED LEARNING-SCIENCE REFERENCE]
[AUTHOR OR SUBJECT-MATTER EXPERT REVIEW REQUIRED]

All sources are authoritative, current, and directly relevant to the learning outcomes of this lesson. Some page numbers and specific references require verification from the published textbook.

👤