How To Use AI Effectively To Enhance Learning

Lately, I’ve been watching how students use AI while they study. The pattern is almost always the same.

A student types something perfectly reasonable like “Explain this topic” or “Improve my paragraph”. AI replies with something that sounds fluent, yet the student may still feels stuck. The answer is long, but it does not quite move them forward.

I don’t think this happens because students are “bad at prompting”. I think it happens because AI is not a teacher. It doesn’t instinctively know your goal, your level, what your teacher expects, or what you already understand. When we leave those things unsaid, AI fills in the gaps with assumptions, and the output becomes generic.

What I’ve found is that the quality of AI support often depends less on the tool itself, and more on whether the student is able to describe the task clearly.

What Clarity Changes

When a student shares a bit more context, the AI response tends to shift in three noticeable ways.

It becomes more specific, because the tool has a clearer target.It becomes more usable, because it produces an output that matches the task.It becomes more reliable, because it guesses less and checks more.

This matters because “AI for school” is not really about getting an answer. It’s about improving your thinking process. When the output is vague, students either copy it and learn very little, or they abandon it and feel that AI is overrated. Neither outcome helps.

The Mindset I Hope Students Can Adopt

The most helpful shift I’ve seen is when students stop treating AI like a search engine and start treating it like a support tool they can brief.

Not in a robotic way. Not with complicated “prompt engineering”. Just with the kind of clarity you would naturally give a tutor.

• What are you trying to do?

• What level are you at?

• What material are you working from?

• What would a good answer look like?

When those pieces are present, AI tends to support learning rather than replace it.

Three moments where AI becomes genuinely useful

Here are three situations where I’ve consistently seen AI help students, without taking away ownership of their work.

1) When you are trying to understand, not just summariseStudents don’t need more words. They need a simpler explanation, a concrete example, and a quick way to check if they truly understand.

2) When you want feedback, not replacementThis is where the line between help and harm becomes very clear. Used well, AI can highlight what’s missing, what’s unclear, and what to improve next. Used poorly, it becomes a shortcut that slowly erodes voice and confidence.

3) When you want to build something and learn through makingI’ve become increasingly interested in “vibe coding” as a learning pathway. Not because every student needs to become a programmer, but because building small tools forces clarity, iteration, and problem-solving. Those skills transfer.

The Prompts I Keep Coming Back To

When I use AI for learning, I return to a small set of prompts that protect two things at once: clarity and ownership. They are simple, but they work because they force the AI to do a specific job.

• Prompt 1: Explain it, then test me“Explain [topic] at [my level] using this material: [paste notes/text]. Keep it to 6 bullet points, define key terms in one line, and give one example. Then ask me 5 check questions from easy to hard. Don’t give the answers until I respond.”

  
  

• Prompt 2: Active recall, not rereading“Turn this into 12 active recall questions (6 basic, 4 application, 2 higher-order) based only on what I pasted: [paste content]. Don’t show answers first. After I attempt, provide a marking guide and model answers.”

  
  

• Prompt 3: Mark my work like a teacher, not a ghostwriter“Here is the question: [paste]Here is what my teacher looks for / the rubric: [paste]Here is my answer: [paste]Give me 3 strengths and 3 improvements. Point to the exact lines that are unclear or off-point. Then show one improved paragraph using my original ideas and examples, and explain what changed and why.”

  
  

• Prompt 4: Stop guessing and ask me first“Before you answer, list the assumptions you are making. If any assumption could be wrong, ask me up to 3 clarifying questions first.”

  
  

• Prompt 5: Vibe coding, but keep it small and workable“I want to build a simple [tool] that does [one main function]. I’m a beginner. Propose the smallest working version (MVP) first, list the steps, then give the code in small chunks with brief explanations. After each step, tell me how to test it. If it fails, ask me to paste the error and debug with me.”

If you try only one, start with the first. “Explain it, then test me” is where AI stops being a content generator and starts becoming genuinely useful for learning.

Where I Stand On Integrity

The biggest risk is quietly losing skill.

If AI is doing the thinking you were meant to do, it’s not support anymore. It’s substitution. And that comes with a cost, even when nobody sees it.

But if AI is helping you practise, reflect, tighten your reasoning, and improve your work with intention, then it is doing what it should do: serving students.

A Question I’ve Been Thinking About

If AI is going to be part of education long-term, then the real advantage won’t belong to the student who uses AI the most.

It will belong to the student who uses AI with the most clarity and the strongest ownership.

---

That Geography Teacher's Game Plan (Physical Copy)

Buy Now

That Geography Teacher's Game Plan (e-book)

Buy Now

If you want to master essential skills, learn clear strategies, and gain confidence for your exams, consider this guidebook. I have distilled over a decade of teaching experience into practical advice tailored just for you.

You may also click on https://www.thatgeographyteacher.com/category/all-products to read about how you can use this guidebook effectively to enhance your learning of Geography.

For additional support to enhance your learning, head to

🔗 https://www.thatgeographyteacher.com/newsyllabus 

You’ll find sample answers to both the 2024 and 2025 O and N-Level national exams. These are ideal for applying the techniques taught in Chapter 2, especially for understanding what a top-band LDQ or well-scaffolded structured answer looks like in reality. You can attempt the questions using the frameworks in the guidebook, then compare against the samples to learn from real answers.