🤖 UNLIMITED POW...QUESTIONS!

📚 Knowledge builders

• What does an ITS need to do? → Here is my list of what an ITS needs to do for it to stand a chance at actually improving student learning.

  1.  🧠 Comprehensive Curriculum by Experts To ensure effectiveness, AI tutoring systems need a curriculum crafted by human experts. This ensures depth, accuracy, and relevance in the material, providing a solid foundation for learners. For example, a math curriculum designed by experienced educators can include real-world applications to make abstract concepts more relatable.

  2. 🏗️ Scaffolded Learning with Gradual Hints: Instead of giving direct answers, AI tutors should offer hints that guide learners through problem-solving. This scaffolding approach supports learners just enough to push them towards deeper understanding. For instance, in a coding exercise, the AI could provide hints about syntax errors or logic flaws without giving away the solution.

  3.  🔄 Active Recall & Spaced Repetition: Incorporating active recall and spaced repetition helps reinforce long-term memory. Learners revisit material at increasing intervals, ensuring information sticks and reducing cognitive overload. For example, an AI tutor could quiz students on vocabulary words at spaced intervals to ensure retention over time.

  4.  🔀 Interleaving: AI tutors should incorporate interleaving, where different types of problems or subjects are mixed together rather than being studied in blocks. This approach helps learners distinguish between concepts and apply them more flexibly. For example, mixing math problems on algebra, geometry, and statistics in a single session can improve overall understanding and retention.
     

  5. 📝 Problem Generation: AI tutors should encourage learners to create their own problems. This generative learning approach deepens engagement by having learners synthesize information and make connections between concepts. For instance, after learning a new scientific principle, students could be asked to design their own experiments to test the principle.
     

  6. ❌ Error-Correction & Reflective Feedback: Rather than just marking answers right or wrong, AI systems should provide feedback that helps learners understand their mistakes. This promotes critical thinking and helps learners correct misconceptions. For example, in a history lesson, if a student gets a date wrong, the AI could provide context about the event to help them remember it better.

  7. 🧩 Cognitive Load Management: Effective AI tutors design interactions to avoid overwhelming learners. Breaking down complex tasks and providing clear instructions helps learners focus on problem-solving without unnecessary distractions. This includes carefully thinking about the user interface of the ITS as this is often the first barrier to pupil success.

What do you think of this list? Is there anything you think I have missed? Let me know in the comments or by replying to this email!

🤖 Industry updates

• Chat GPT 4.5 Preview → GPT-4.5, OpenAI's latest model, enhances its ability to recognise patterns, draw connections, and generate creative insights, making it a valuable tool for improving writing and solving practical problems. The model's broader knowledge base and improved user intent following can help teachers create more engaging and effective educational content. Additionally, GPT-4.5's reduced hallucinations and increased reliability make it a more trustworthy assistant for educational purposes.

✨ Fresh prompts

1. Dynamic Examples → The dream of every teacher is to make sure you can have infinite examples of the specific skill that you are teaching pupils, right? With Claude’s artefacts tool, this is possible. Dream it, then describe it and after a little back and forth you have a tool that can do just that.

You are an expert at creating dynamic worksheets where questions can be changed with the touch of a button. You want this to be able to model to children with infinite examples. You also want a sentence stem to be present that reads, "This is a ___ angle because it is ___ than 90 degrees." You understand the importance of small steps, so your dynamic questions always fulfill a strict criteria that is never broken. Create a dynamic worksheet that allows pupils to identify acute, right, and obtuse angles only. There should be no reflex angles. The examples should appear in many different orientations too. Make sure the regeneration button works and does not use reflex angles in the examples. Keep all angles less than 180 degrees. I would also like a working button to toggle on or off the angles in the question.

You will most likely need a bit of back-and-forth between yourself and Claude to really nail what you want.

Play around with an example of the dynamic example generator here.

Got an example of some interesting artefacts you have made with Claude? Share it with us, by clicking ‘Publish’ in the bottom right corner of the Artifact screen.