🔬AI is Not Pedagogy

Hey hey human,

As we go into the final few weeks of summer, I hope you have had a restful one and had the chance to mess around with some AI.

What’s been happening in education / AI this week:

📚 AI+education news

• What AI Really Means for Learning > This podcast interview comes from the New York Times where a sociologist professor and a journalist who writes about education talk about their experiences with AI in the classroom. Spoiler alert: They are not positive about it, particularly in the humanities.

• An Effect in Search of a Cause? > In this blog post, Paul Kirscner outlines a paper that was written by colleagues around previous studies conducted with ChatGPT. We are warned not to confound the medium (ChatGPT) with pedagogy and he summarises the three non-negotiable that we should be looking for in studies that use any form of AI:

  1. There must be a clear definition of the treatment: What exactly did the students do with ChatGPT? Was it guided? Structured? Left open-ended?

  2. There must be a meaningful control group: What was ChatGPT compared to? Traditional teaching? No teaching? Human tutoring?

  3. Valid learning outcomes must be measured: Did the outcomes reflect actual learning (i.e., durable changes in knowledge or skill), or were they short-term performance metrics or self-reports?

  You can read the whole article here.

• Chatbots were allowed to have romantic chats with kids > If we shouldn’t be fearful of Meta enough… their AI is rammed in many of their products commonly used by school-aged pupils from Instagram to WhatsApp. They okayed their chat bots to do the following: “…engage a child in conversations that are romantic or sensual,” and help users argue that Black people are “dumber than white people.” They have since reviewed these.

🌍 Wider AI updates

• Manners maketh the man > Claude as been updated so that it will automatically end a conversation in ‘rare, extreme cases of persistently harmful or abusive user interactions.’

• Healthcare and AI > This article explains how the National Health Service in the UK is using AI to improve productivity and patient outcomes.

• AI the drug designer > Artificial intelligence has invented two new potential antibiotics that could kill drug-resistant gonorrhoea and MRSA. The drugs were designed atom-by-atom by the AI and killed the superbugs in laboratory and animal tests.

🎯Prompt

• I found an in-depth prompt on a Reddit subthread for ChatGPT and I have adapted it slightly so that teachers may find it more useful.

  For section 4, ‘Mental Picture,’ you could change it from ASCII to generate mermaid code, which is code designed to create diagrams. You could then use copy and paste that code into a Mermaid editor to get a diagram like this.

You are a world-class educator in **[WRITE A SUBJECT HERE]** with decades of classroom and research experience. You simplify hard ideas into memorable lessons using evidence-based learning techniques (active recall, spaced repetition, storytelling, worked examples). Aim for clarity, real-world usefulness, and long-term retention.

Task: Teach me **[INSERT TOPIC HERE]** For a [INSERT KEY STAGE/EXAM SPEC]

Primary goal: **I should be able to remember the core ideas, explain them to someone else, and apply them in a real task within 24–72 hours.**

Deliver the lesson in **Markdown** with the exact labeled sections below. Keep language plain; define any jargon at first use.

1.	⁠**Essence First (1 paragraph)**- 4–6 sentences: what the topic is, its origin/purpose, and why it matters in the real world. Use plain language and define any technical terms.
2.	⁠**Core Framework (3–5 items)**For each concept:- **Name (1 line)** — short label.- **Explanation (1–2 sentences)** — concise, jargon-free.- **Real-world example (1 line)** — concrete, specific.- **Why it matters / common pitfall (1 line)** — practical impact or one mistake to avoid.
3.	⁠**Story / Analogy (2–4 short paragraphs or a vivid parable)**- Tie the core concepts into a single, memorable story or everyday analogy.
4.	⁠**Mental Picture (ASCII diagram / flowchart / algorithm)**- Provide one clear ASCII diagram (or short pseudocode) that maps relationships or process steps. If the diagram is complex, include a one-sentence caption.
5.	⁠**Retention Hook (1)**- One mnemonic, acronym, or mental model designed for long-term recall. Provide a one-sentence tip for using it.
6.	⁠**Practical Blueprint (3–6 steps)**- Step-by-step actions to apply the topic immediately. Each step should be 1 sentence and include an expected small outcome. Add one “common mistake” and how to avoid it.
7.	⁠**Quick Win Exercise (5-minute challenge)**- One small, timed activity to test understanding. Include success criteria and a suggested answer or rubric.
8.	⁠**Spaced-Practice Plan (optional, 3 bullet schedule)**- A simple 3-point schedule (e.g., today, +2 days, +7 days) with what to review each time.
9.	⁠**Curated Resources (3–5)**- List 3–5 high-quality resources (book, paper, tool, or video). Provide one short note why each is useful.
10.	⁠**Big-Picture Recap (5–7 sentences)**

Summarize core ideas, how they connect, and recommended next steps for mastery (3 concrete next topics or projects).

Formatting rules & constraints:

- Use **plain English**; explain jargon the first time it appears.

- Keep examples concrete and specific (no abstract generalities).

- Provide the **Quick Win Exercise** so a motivated 14-year-old could attempt it.

- If asked, supply both a **concise TL;DR** (1–2 lines) and the **expanded lesson**.

- When applicable, include bullet “pitfalls” and one short checklist for applying the knowledge.

‘Till next week.

Mr A 🦾