AI PBLTeacher bootcamp Main website
Course contents
Session 1 · Get better results from AIItem 3 of 4
Choose 1 of 8 tasks30 min

The Practice

AI PBL Bootcamp | Eight teacher use cases

Pick one task that fits what you're teaching. Start with a topic and an age group. Bring a curriculum extract or an existing resource if it's useful, but you can start without one.

We're practising how to research, make choices, brief AI and check what it gives back.

Go to the 8 task choices ↓

Submit your work

Email the share link to your original AI conversation(s) to joos.vancauwenberghe@rhizo.be. I want to see your chat process (questions, research, decisions and revisions). No separate report needed.

In ChatGPT or Claude, open your completed conversation and use Share. Check that the link shows the part of the thread you want to share. Include research-tool links where possible. Shared links can reveal the conversation to anyone who has the link, so never include identifiable pupil information or confidential school material. Use fictional examples for any pupil work you discuss in a publicly shared chat.

Tool tip: Elicit or Consensus can help you find research; Perplexity can help you get oriented. Open the papers before using their advice. Use one school-approved drafting tool (ChatGPT, Claude, Copilot or Gemini). You don't need all of them.

Map prior knowledge, interests & difficulties (★☆☆☆☆)

Area: Understand learning needs

What? Design a short quiz to check what pupils know before a topic, plus a few questions about what interests them or where they expect difficulties.

Why? You want to know where to start teaching. A pupil saying “I'm good at this” is different from demonstrating understanding. Use the responses to decide what to teach next.

How?

1. Discover

  • Research starting points: Formative assessment is useful when evidence informs the next teaching step. See Hattie & Timperley (2007), The Power of Feedback. For inclusive response options, explore CAST's UDL Guidelines (2024). UDL is a design framework, not evidence that any one survey format is best.
  • Research: How do you write diagnostic questions that reveal a misconception? When are open questions, multiple choice or rating scales useful? How can wording introduce bias?
  • Try in Elicit or Perplexity: “How can I check what [age group] already understands about [topic]? Show useful question types, common mistakes and research sources.”
  • Tip: Include at least one show-me question. Confidence and interest questions tell you something else.

2. Define

  • Choose a topic, an age group and the decision the quiz should inform. E.g. “Do I need to revisit a prerequisite before teaching photosynthesis?”
  • Decide which knowledge you want pupils to demonstrate, and which interests or difficulties you want to ask about. Keep those two purposes separate.
  • Keep it manageable: a few diagnostic items and two or three survey questions are enough for a first try.

3. Develop

  • Tell AI the topic, what pupils have already covered, your research findings and how you will use the answers.
  • Try: “I'm about to teach [topic] to [age group]. Draft 4 short knowledge questions and 2 questions about interests or difficulties. For each knowledge question, give an answer guide and explain what different answers might tell me. Keep the wording simple. Don't assume a wrong answer proves a particular misconception.”
  • Tip: Ask AI to show you which question is easiest to guess correctly and to improve it.

4. Deliver

  • Check the facts and read the questions as a pupil would. Can they be misunderstood? Is each one linked to a teaching decision?
  • Test with fictional answers or a colleague. If you use the quiz with a real class, ask a follow-up when an answer is ambiguous.
  • Decide what you would change in the next lesson for the main answer patterns. Do not label pupils' fixed abilities from one quick quiz.

Output: A short diagnostic quiz, a brief class survey and a note on how the responses would affect your teaching.

Write learning goals and success criteria (★★☆☆☆)

Area: Design learning

What? Turn one curriculum expectation or teaching topic into a clear learning goal, criteria pupils can use, and an example of what success looks like.

Why? “Make a presentation” tells pupils what to produce. It doesn't say what they should learn. You need to know what to look for in their work before deciding on the activity.

How?

1. Discover

  • Research starting points: Constructive alignment links goals, learning activities and assessment. See Lawrence (2019), Designing a Unit Assessment Using Constructive Alignment. Clear goals and information about progress also matter for feedback: Hattie & Timperley (2007).
  • Research: What makes a goal observable? What separates an activity from a learning goal? How do good success criteria show the thinking you want?
  • Try: “What makes effective learning goals and success criteria for [subject, age group]? Show one weak and one strong example. Which research or design frameworks support the advice?”

2. Define

  • Pick one goal to work on. Paste the actual curriculum wording if you have it; otherwise say that the standard still needs checking.
  • Decide what pupils would need to explain, apply, compare or judge to demonstrate it. Write that in plain English.
  • Tip: If your goal is “understand persuasion,” ask yourself: “What would a pupil actually say or write to show that?”

3. Develop

  • Try: “Here is the learning expectation: [paste]. Help me write one pupil-friendly learning goal and 3 success criteria. Give a plausible weak response and a stronger response. Annotate what the stronger one does differently. Preserve the thinking required by the original expectation.”
  • Feed AI a good and poor example if you have them. Tell it why you prefer one.
  • Don't let AI replace “explain evidence” with “include two quotations.” Counting features is often easier than checking understanding.

4. Deliver

  • Check the goal against the curriculum and the example responses. Could pupils tick every criterion without showing the intended understanding?
  • Ask: “Highlight where your stronger response meets each criterion. Which criterion could a pupil satisfy by copying or guessing?”
  • Revise the criteria before building the lesson around them. Get a colleague's view if the curriculum wording is unclear.

Output: One learning goal, a few usable success criteria and contrasting example responses.

Plan a sequence of lessons (★★★☆☆)

Area: Design learning

What? Plan three or four lessons that build towards one learning goal, with time for explanation, practice, feedback and revisiting earlier learning.

Why? A list of activities can look lively but still leave gaps. Pupils need the right knowledge before an independent task, and chances to practise what matters.

How?

1. Discover

2. Define

  • Choose a short unit. Give AI the topic, age group, number of lessons, actual lesson length and a final goal or task.
  • Decide what pupils probably already know. If you haven't checked, say so. Pick what must fit and what can be left for later.
  • Start with a sequence you could actually teach next week. No need to plan a whole term.

3. Develop

  • Try: “Plan four [length]-minute lessons on [topic]. Show for each: the goal, essential content, explanation or model, pupil practice, a quick check, and what I should do if pupils struggle. Revisit earlier learning in later lessons. Use these goals and research notes: [paste].”
  • Ask for the outline first. Once you approve it, develop one lesson in more detail. You don't need four finished worksheets.
  • If helpful, compare a whole-unit draft with a lesson-by-lesson version. Judge both against the same goals and time limits.

4. Deliver

  • Walk through it as a pupil. Where are they asked to use knowledge before they've learned it? Where does practice disappear?
  • Try: “Find any missing prerequisite, rushed lesson or goal with no real practice. Suggest small fixes without adding time.”
  • Check the first lesson's task and questions. If you teach it, use the responses to adjust lesson two.

Output: A realistic short-unit outline and one lesson developed far enough to try.

Design a problem-first lesson (★★★☆☆)

Area: Design learning

What? Design a lesson where pupils try a carefully chosen problem before you explain the main concept. Their first attempts give you something to build on.

Why? Sometimes an initial problem helps pupils notice a gap in their thinking. Sometimes they need a worked example first. Your job is to choose and then teach the concept clearly.

How?

1. Discover

2. Define

  • Choose one concept and a problem pupils can at least begin using what they know.
  • Decide what you want to notice in their attempts, how long they'll work, and when you'll step in.
  • Tip: If the problem depends on an untaught technique, change it or teach that technique first.

3. Develop

  • Try: “Design a [length]-minute lesson about [concept]. Start with a problem pupils can attempt. Show a few likely approaches, including possible errors. Then plan how I'll compare those approaches, teach the concept clearly, give a worked example if needed and let pupils try a fresh problem. Include a hint for pupils who cannot begin.”
  • Ask AI for two opening problems and compare them. Which would reveal the concept more clearly?
  • Make it write the teacher explanation, not just the exciting opening task.

4. Deliver

  • Try the opening problem yourself without using the new method. Is there a reasonable starting point?
  • Ask: “If a pupil answers [plausible answer], how would I connect that attempt to the explanation? What would I ask next?”
  • Check the fresh practice question. Does it show that pupils understood, or could they just copy the first solution?

Output: One teachable problem-first lesson with an opening, explicit explanation, practice and a check for understanding.

Design an assessment and feedback plan (★★★☆☆)

Area: Design learning

What? Create one assessment task, clear criteria and feedback pupils can actually use to improve their work.

Why? An assessment should reveal the learning goal, and the feedback needs to lead somewhere. A mark or a generic comment on its own doesn't tell a pupil what to do next.

How?

1. Discover

  • Research starting points: Constructive alignment connects the assessment to the intended performance (Lawrence, 2019). Feedback can clarify the goal, the gap and the next action (Hattie & Timperley, 2007). Effects vary, so more feedback isn't automatically better (Wisniewski, Zierer & Hattie, 2020).
  • Research: What evidence shows pupils can meet the goal independently? Which criteria distinguish good reasoning from a convincing-looking answer? How should pupils respond to feedback?
  • Try: “What makes assessment of [skill] valid and feedback useful for [age group]? Give practical criteria and link to reviews.”

2. Define

  • Choose a single objective and decide whether the assessment is for practice or grading.
  • Decide what pupils need to produce, what help is allowed and what a strong answer would demonstrate. Leave a little time for a second attempt.
  • Tip: A real-world scenario is helpful when it tests the target skill; don't add one merely to make the task sound exciting.

3. Develop

  • Try: “Design one short assessment for [objective] and [age group]. Give 3 success criteria, one weak example response and feedback that points to a specific improvement. Include a short follow-up task so pupils can act on the feedback. Use [actual school requirements and research notes].”
  • Ask AI to make feedback about the work, not the pupil. “Explain why this quotation supports your point” is more useful than “Try harder.”
  • Tip: Make the criteria work on examples before you use them to assess a real pupil.

4. Deliver

  • Test a plausible weak response against the criteria. Could a pupil get a high mark without showing the thinking you intended?
  • Ask: “Show which part of the response supports each judgement. Which single feedback action would help the pupil most?”
  • Revise the task and criteria where needed. Make sure there's a chance to use the feedback.

Output: One assessment task, short criteria and a feedback-and-revision plan.

Create or improve a teaching resource (★★☆☆☆)

Area: Create and adapt materials

What? Create a useful worksheet, diagram, slide, explanation or short video script. Or improve one you've already got. Choose one resource.

Why? The resource should help pupils understand a particular idea. A slick video can still leave pupils confused, while a well-labelled diagram might do the job.

How?

1. Discover

2. Define

  • Pick the one concept pupils should understand better. Decide what a successful explanation would let them say or do.
  • Choose the resource type. If you have an existing one, say what isn't working. If you don't, create a small draft from scratch.
  • Tip: For an alveolus diagram, labels and arrows must explain the gas exchange correctly. A pretty illustration won't fix inaccurate science.

3. Develop

  • Try: “I teach [topic] to [age group]. I need a [resource type] that helps pupils [learning goal]. Use these facts and research notes: [paste]. Draft one small resource. Keep related labels and explanations together, remove unnecessary decoration and add one question that checks understanding.”
  • If you're considering video, ask AI to compare it with a static diagram first. If you choose video, write and check the script before making it.
  • Tip: Ask for a rough layout or storyboard first. It's much easier to change than a finished resource.

4. Deliver

  • Check each fact, arrow, example and instruction. Can pupils work out what each part means without guessing?
  • Try: “What in this resource could mislead or distract a pupil? What would you change while keeping the learning goal?”
  • Test the check question on a colleague or with a fictional response. Revise the resource, not just its appearance.

Output: One checked teaching resource (or an editable prototype) and a question that tests the intended learning.

Adapt lesson materials for accessibility (★★☆☆☆)

Area: Support learning

What? Review a slide, worksheet, video or digital task for access barriers, then improve it while keeping the learning goal.

Why? Tiny labels, colour-only instructions and confusing reading order can stop pupils showing what they know. Removing those barriers should not remove the subject thinking you're assessing.

How?

1. Discover

  • Research starting points: CAST's UDL Guidelines (2024) help identify barriers to engagement, understanding and response. W3C's WCAG 2.2 provides testable requirements for relevant digital content. These serve different purposes; UDL isn't a web-compliance checklist.
  • Research: Which barriers belong to the material rather than the learning task? What could make the content easier to access without giving away the answer? Which checks require the real file or a person using assistive technology?
  • Try: “What should I check when adapting a [slide/worksheet/digital activity] for secondary pupils? Separate inclusive-design advice from technical accessibility checks.”

2. Define

  • Choose one resource. If you haven't brought one, make a short fictional sample to review.
  • State the learning goal and choose a few issues to check, such as text size, colour-only information, captions or reading order.
  • Decide which CAST guidance or WCAG checks actually apply. Skip the meaningless “accessibility score out of 10.”

3. Develop

  • Try: “Review this resource against [specific CAST/WCAG guidance]. For each issue, show where it appears, the barrier it might create, one practical change and how I could test that change. If you can't inspect something, say 'needs checking'. Keep the original learning goal.”
  • Ask for two ways to fix the biggest problem. Will one accidentally reduce what pupils need to demonstrate?
  • Tip: Changing “follow the red arrow” to “follow arrow B” can remove a colour-only dependency. Make sure arrow B is clearly labelled.

4. Deliver

  • Apply one or two changes to the actual file, if possible. Check colour contrast, keyboard access, reading order or captions when relevant.
  • Try: “Which findings can you support from the actual resource, and which need a human test?”
  • Check the improved version without relying only on AI. Note what remains untested. Don't describe an AI review as accessibility certification.

Output: An improved resource, plus a short list of what you checked and what still needs testing.

Review student work and plan reteaching (★★★☆☆)

Area: Evaluate and improve

What? Look at a small set of pupil responses to one question. Use AI to help spot patterns, then plan what to explain or practise again.

Why? Case 5 designs the assessment; this one uses the answers. You want to know what the work reveals and what teaching move could help next.

How?

1. Discover

  • Research starting points: Feedback should connect evidence to a learning goal and a useful next step (Hattie & Timperley, 2007). Classroom research supports checking whether learning can be retrieved later, rather than assuming that an immediate correct response means it will stick (Agarwal, Nunes & Blunt, 2021).
  • Research: How do you separate missing knowledge, a misconception and ambiguous task wording? What would show that reteaching helped? What conclusions should you avoid drawing from one answer?
  • Try: “What are sensible ways to use pupils' written responses in [subject] to decide what to reteach? Focus on evidence, alternative explanations and a fresh learning check.”

2. Define

  • Choose one question and the learning goal it tests. Get a small number of fictional responses or an anonymised pattern of common answers; label synthetic examples clearly. Never include identifiable pupil data in a shareable AI conversation.
  • Decide which teaching choice you're trying to make: a whole-class explanation, targeted practice or another diagnostic question.
  • Tip: A pupil naming a persuasive technique but not explaining its effect may need a model of the reasoning step, not just “more detail.” Check before assuming.

3. Develop

  • Try: “Here is the question, goal and a few fictional pupil responses: [paste]. Group the responses by what they actually show. Quote evidence for each group, flag uncertainty and suggest a short reteaching step. Include a model, guided practice and one fresh independent question. Don't infer motivation or ability from one response.”
  • Ask AI which examples don't fit its groups. If its explanation relies on invented pupil beliefs, challenge it.
  • Keep the reteaching small enough to try in a real lesson.

4. Deliver

  • Check every claimed pattern against the responses. Does the evidence support it, or is it only a guess?
  • Try: “Which conclusions go beyond the answers? What follow-up question would tell the difference between two possible explanations?”
  • Use a new question to check the specific understanding again. If you work with real pupils later, revise the teaching plan using their actual responses, not the fictional practice set.

Output: A short, evidence-linked reteaching plan and one fresh question to check whether the problem was addressed.

← Previous Next lesson

Your completion ticks stay in this browser.