Week 8: AI as a Study Partner

 

Welcome to Week 8 of the AI Literacy Course.

In Lesson 7, we learned how to evaluate an AI response and decide whether to follow up, verify, restart, or stop.

This week, we ask a practical question:

How can AI support learning without hiding who performed the learning?

AI can explain concepts, ask practice questions, provide hints, generate examples, offer feedback, and help organise study. These capabilities can be useful, but completing a task is not automatically the same as learning.

A good study partner should increase the learner’s understanding, participation, and control.

Learning goals

By the end of this lesson, learners should be able to:

• Distinguish task performance from learning

• Define an observable learning goal

• Make an initial attempt or identify a specific difficulty

• Choose the smallest useful form of AI support

• Use retrieval practice and graduated hints

• Request feedback without surrendering authorship

• Check course rules, important content, and privacy

• Demonstrate progress through a parallel task and reflection

1. Task performance is not always learning

Performance means completing the task in front of you.

Learning means developing knowledge or skill that you can later explain, recall, adapt, judge, or apply.

Imagine that a learner asks AI to write a complete explanation of climate change. The resulting text may be clear and accurate. It demonstrates what the AI can produce, but it does not prove that the learner understands climate change.

Now imagine that the learner:

• writes an initial explanation;

• asks AI to identify one unclear point;

• studies that point;

• revises the explanation;

• answers a new question without AI supplying the answer.

This process provides stronger evidence of learning.

The goal is not merely to produce a polished result. The goal is to build understanding and control.

2. Begin with a visible learning goal

A study interaction should begin with what the learner wants to understand, practise, remember, or produce.

Vague goal:

“Help me with statistics.”

More visible goal:

“I need to distinguish correlation from causation and apply both concepts to new examples.”

The second goal can be demonstrated.

Useful learning goals may begin with:

• “I want to explain…”

• “I want to compare…”

• “I want to solve…”

• “I want to remember…”

• “I want to apply…”

• “I want to identify…”

• “I want to evaluate…”

A visible learning goal helps determine what kind of AI support is appropriate. It also provides a basis for checking whether learning has occurred.

3. Make an initial attempt or identify the difficulty

When possible, think before asking AI to provide an answer.

An initial attempt might involve:

• solving the first step of a problem;

• writing a short explanation;

• recalling what you know without notes;

• creating your own example;

• identifying where you became stuck;

• predicting which method might work.

The attempt does not need to be complete or correct. Its purpose is to reveal current understanding.

For example:

“I think correlation means that two things change together, but I am not sure why this does not prove causation.”

This gives AI a specific difficulty to address.

When a learner cannot begin

“Attempt first” should not become a rigid rule.

A complete beginner may need:

• a definition;

• a short explanation;

• one worked example;

• a choice between possible first steps;

• help identifying what knowledge is missing.

In that situation, the learner can ask for limited introductory support and then attempt a similar task.

A useful principle is:

Think first when possible, then request the smallest support needed to continue.

4. Choose the smallest useful AI role

AI can support learning in different ways. The best role depends on the learning goal and the learner’s current difficulty.

Possible forms of support include:

• explaining one concept;

• asking diagnostic questions;

• giving a hint;

• generating a practice question;

• showing one example;

• creating a new variation;

• providing feedback against criteria;

• helping organise a study session.

Before prompting, ask:

What is the smallest useful help that allows me to continue thinking?

If a learner is stuck on one step of a mathematics problem, the smallest useful help may be a guiding question—not a complete solution.

If a learner lacks the necessary background knowledge, a short explanation may be appropriate before practice begins.

The level of support can increase gradually:

1. Ask a guiding question.

2. Name the relevant concept.

3. Give a small hint.

4. Show one partial step.

5. Demonstrate a similar example.

6. Reveal the full solution only when necessary, followed by a new learner attempt.

This is called graduated support.

When the learner is ready, the support should decrease. This is called fading support.

5. Practise actively

AI study support is most useful when the learner must think, recall, explain, or apply.

Retrieval practice

Retrieval practice means trying to recall information before seeing the answer.

Example prompt:

“Ask me one question at a time about photosynthesis. Wait for my answer. Then tell me what was correct and give one hint if something important is missing.”

Retrieval practice helps reveal what the learner can recall and where more study is needed.

AI-generated questions still require review. They may focus on unimportant details or include incorrect assumptions.

Generate your own examples

After seeing one example, ask AI to let you create the next one.

Example:

“Give one example of correlation without causation. Then ask me to create a different example and give feedback on it.”

Creating an example requires more active understanding than simply reading one.

Practise transfer

Transfer means applying the same idea in a new context.

For example, after learning a concept through a school example, the learner might apply it to:

• a workplace situation;

• personal finance;

• public health;

• civic life;

• a scientific question.

The concept should remain stable while the context changes.

Generated scenarios may contain stereotypes, unrealistic details, or factual mistakes. The learner should examine them before use.

Turn explanations into activity

An explanation can be followed by a learner action:

“Explain the concept in three short parts. After each part, ask me to summarise the main idea in my own words.”

Lesson 9 will explore question-based and Socratic dialogue in greater depth.

6. Use feedback without surrendering authorship

AI can provide feedback on a learner’s attempt, but useful feedback should support revision rather than replace the learner’s work.

A useful pattern is:

“Using the assignment criteria below, identify one strength, one question, and one priority for improvement. Quote the relevant part of my answer. Do not rewrite it for me.”

This keeps the learner involved.

Effective feedback should:

• refer to evidence in the learner’s work;

• connect to visible criteria;

• identify a manageable priority;

• explain why the issue matters;

• leave the revision to the learner.

AI does not automatically know the teacher’s expectations, course outcomes, or assessment practices. Provide the relevant criteria when this is allowed.

AI feedback may also be inconsistent, overly positive, or based on incorrect assumptions. It is advisory and does not replace teacher assessment.

Preserve the learner’s voice

AI may help identify unclear language, but complete rewriting can erase the learner’s style and hide learning needs.

A better prompt is:

“Identify two sentences that are difficult to understand and explain why. Do not rewrite the paragraph.”

The learner should understand and be able to explain every final sentence.

Reading and language support will be developed further in Lesson 10.

7. Check rules, content, and privacy

Before using AI for schoolwork, check the rules that apply.

Ask:

• Is AI use permitted for this task?

• Which kinds of assistance are allowed?

• Must AI use be disclosed?

• Which parts must be completed independently?

• May the material be uploaded to an AI tool?

• Is there an approved tool?

Different teachers, schools, courses, and assignments may have different rules.

Check important content

AI-generated explanations, examples, calculations, questions, and feedback may contain errors.

Compare important content with:

• course material;

• a textbook;

• teacher guidance;

• an appropriate authoritative source;

• a reliable calculation method.

Asking AI “Are you sure?” is not independent verification.

Protect privacy

Do not upload confidential, sensitive, or identifying information merely because it may improve the response.

Use:

• the minimum information needed;

• fictional examples;

• general descriptions;

• approved tools.

Serious personal, medical, legal, crisis, or safeguarding concerns require accountable human support.

8. Reduce support and test learning

After receiving support, try a similar task with less AI assistance.

This is a parallel task: a new task that uses the same skill or concept.

For example:

• First task: distinguish correlation and causation in a health example.

• Parallel task: apply the same distinction to an education example.

The parallel task should be similar enough to test the same learning but different enough to require a new response.

Ask:

• Can I explain the central idea in my own words?

• Can I apply it to a new example?

• Can I identify and correct an error?

• Can I explain why the answer works?

• Can I complete the task with less AI help?

• Can I judge whether an AI answer is suitable?

This is stronger evidence of learning than a polished AI-assisted output.

Accessibility support may continue

Reduced AI support does not mean removing every tool or accommodation.

A learner may appropriately continue using:

• a screen reader;

• speech-to-text;

• translation support permitted by the course;

• spelling assistance;

• adapted formats;

• other approved accessibility tools.

The purpose of the parallel task is not to remove fair support. It is to check that AI is not supplying the knowledge or answer being assessed.

Healthy use means informed control, not complete independence from every tool.

9. Reflect on agency and progress

After the activity, consider what changed.

Ask:

• What could I do before using AI?

• What kind of support did I request?

• What did I do myself?

• Which AI suggestion did I accept?

• Which suggestion did I change or reject?

• What can I now explain or apply?

• Where do I still need support?

• Did AI strengthen my thinking or replace part of it?

• Was my use consistent with the course rules?

AI use becomes problematic when the learner cannot explain the final work, depends on repeated reassurance, or hides how the result was produced.

Appropriate AI use should make the learning process more visible—not less.

Classroom activity: Study-partner experiment

Choose one concept or skill from a course you are studying.

Step 1: Define the learning goal

Complete this sentence:

“After this activity, I want to be able to…”

Make the goal observable.

Step 2: Make an initial attempt

Write, solve, recall, explain, or apply the concept before asking AI for substantial help.

If you cannot begin, explain what is blocking you.

Step 3: Choose one or two support methods

Choose from:

• a short explanation;

• retrieval questions;

• graduated hints;

• one worked example;

• a new practice example;

• criteria-based feedback;

• help organising a study session.

Explain why this support fits your learning goal.

Step 4: Use AI and record what happened

Save the important prompts and responses.

Do not include unnecessary personal information.

Step 5: Revise or practise

Use the AI support, then complete the meaningful thinking or revision yourself.

Step 6: Complete a parallel task

Try a similar task with reduced AI support and any normally permitted accommodations.

Step 7: Compare the stages

Use this study log:

Stage	What happened?	What did I do?	What did AI do?
Learning goal
Initial attempt
AI support
Learner revision or practice
Parallel task
Reflection

Step 8: Write a reflection

In approximately 300–500 words, explain:

• what you wanted to learn;

• what your initial attempt showed;

• which AI support you selected;

• what you did yourself;

• one AI suggestion you accepted;

• one suggestion you changed or rejected;

• what the parallel task showed;

• what support you may still need.

Reflection questions

1. What is the difference between performance and learning?

2. What makes a learning goal observable?

3. Why is an initial attempt useful?

4. What can a beginner do when they cannot make a full attempt?

5. What is the smallest useful support?

6. What is retrieval practice?

7. How do graduated hints protect learner thinking?

8. Why should AI feedback be connected to supplied criteria?

9. How can a parallel task provide evidence of learning?

10. Why is continued accessibility support not automatically unhealthy dependence?

Key vocabulary

Study partner:
AI used to support explanation, practice, feedback, planning, and reflection without hiding who performed the learning.

Performance:
Completing the current task, possibly with assistance.

Learning:
Developing knowledge or skill that can later be explained, recalled, judged, adapted, or applied.

Learning goal:
A visible statement of what the learner should understand, practise, remember, or produce.

Initial attempt:
The learner’s own first effort or description of the difficulty before substantial AI support.

Retrieval practice:
Trying to recall information before seeing the answer.

Graduated support:
Increasing help step by step, beginning with the smallest useful assistance.

Fading support:
Reducing assistance as the learner gains understanding and control.

Transfer:
Applying a concept or skill in a new situation.

Criteria-based feedback:
Feedback connected to visible standards and evidence in the learner’s work.

Parallel task:
A new task that tests the same concept or skill in a different example or situation.

Learner agency:
The learner’s ability to set goals, make choices, evaluate support, and remain responsible for the work.

Summary

In Week 8, we learned that completing a task is not automatically the same as learning.

AI can support explanations, retrieval practice, graduated hints, examples, feedback, and study planning. The learner should still define the goal, think actively, make decisions, and understand the final work.

A responsible study cycle moves through:

Learning goal → initial attempt → smallest useful support → active practice → reduced support → parallel task → reflection

The aim is not to remove every tool. Appropriate accessibility support may continue.

The aim is genuine learning: greater understanding, participation, and control, with transparent AI assistance and meaningful human judgement.

Lesson 8 Interactive Quiz: AI as a Study Partner

Choose one answer for each question. Then select Check my answers. You will receive feedback for every question. This practice quiz does not collect names or scores.

1. A learner submits an excellent AI-written explanation but cannot explain its main argument. What does this demonstrate?

A. The learner has mastered the concept because the text is accurate. B. The learner has demonstrated independent transfer. C. The task was completed, but the text does not prove that learning occurred. D. The learner no longer needs to study the topic.

2. Which is the most observable learning goal?

A. “I want to distinguish correlation from causation and apply both concepts to two examples.” B. “I want AI to make statistics easier for me.” C. “I want to read a long explanation about statistics.” D. “I want a polished answer about correlation.”

3. A beginner does not know how to solve a problem. What is an appropriate initial attempt?

A. Ask AI to complete the entire task without first identifying the difficulty. B. Avoid seeking any support until a complete answer can be produced. C. Copy a worked solution and present it as personal understanding. D. State what is already understood and identify the point where help is needed.

4. A learner is stuck on one step of a mathematics problem. What should the AI study partner try first?

A. Complete the problem and provide only the final answer. B. Ask a guiding question that helps the learner identify the next step. C. Replace the problem with an unrelated, easier task. D. Rewrite the learner’s attempt without explaining what changed.

5. Which activity is the best example of retrieval practice?

A. Reading the same explanation several times. B. Copying AI-generated definitions into notes. C. Answering one question from memory before receiving feedback. D. Looking at the answer before attempting the question.

6. Which sequence best demonstrates graduated support?

A. Guiding question → relevant concept → small hint → partial step → similar example B. Complete solution → guiding question → learner attempt C. Final answer → shorter final answer → longer final answer D. General praise → full rewrite → automatic submission

7. Which prompt best preserves learner authorship?

A. “Rewrite my entire answer so that it meets the criteria.” B. “Replace my wording with a professional version I can submit.” C. “Correct everything and give me the finished assignment.” D. “Using these criteria, identify one strength and one priority for improvement. Do not rewrite my answer.”

8. Before using AI for an assessed task, what should the learner do?

A. Assume the same AI rules apply to every course. B. Check the assignment, teacher, or school rules about permitted assistance and disclosure. C. Use AI freely if the final text is edited afterwards. D. Upload all available course and student information to improve the response.

9. After learning correlation and causation through a health example, which parallel task best tests learning?

A. Copy the original explanation using different formatting. B. Ask AI to answer the same health question again. C. Apply correlation and causation to a new education example with reduced AI help. D. Read the original AI response one more time.

10. A learner continues using an approved screen reader and speech-to-text tool during a parallel task. What does this mean?

A. Appropriate accessibility support may continue without showing unhealthy dependence. B. The task cannot provide evidence of learning because all tools must be removed. C. The learner should stop using approved accommodations to demonstrate independence. D. Any continued technology use means that AI has replaced the learner.

Reflection: Choose one study task. What is the smallest useful AI support that would help you learn while keeping the meaningful thinking and decisions with you?