Last summer, I watched a 13-year-old type "ideas for an essay" into ChatGPT. The AI responded with a generic list of topics in a language she couldn't understand. She closed the tab, frustrated.
This scene repeated itself dozens of times as I built the first AI-integrated Academic Writing course for Year 12-13 and Foundation Year students. The problem wasn't the technology. It was that students had no idea how to communicate with it.
The gap isn't technical. It's instructional.
Students need a framework, not just access. After reviewing recent studies on prompt engineering and testing approaches with hundreds of students, I've found that four simple components consistently improve AI interactions in educational settings:
1. Define the Role
"Act as a Year 12 English teacher reviewing my persuasive essay draft."Research shows that giving AI a clear role helps it understand context and expectations (Walter, 2024). For students, this means better, more relevant feedback aligned to their learning level.
2. Provide the Input
Students must paste their actual work; e.g. the essay paragraph, the thesis statement, the outline. Without specific content, AI defaults to generic responses that don't help anyone learn.
Studies confirm that contextual grounding reduces vague outputs and improves accuracy (Lee & Palmer, 2025).
3. Specify Steps
Instead of "help me," students should break requests into clear actions:"First, identify my thesis statement""Second, check if my evidence supports my argument""Third, suggest one way to strengthen my conclusion"This structured approach mirrors chain-of-thought prompting, which research shows significantly improves the quality of reasoning (Wei et al., 2022).
4. Set Knowledge Level
"Explain at a Year 10 reading level" or "Give feedback as if I'm learning this concept for the first time."
When students specify their level, AI adjusts complexity appropriately. It is a critical factor validated across multiple studies (Lee & Palmer, 2025; Walter, 2024).
What students typically type: "Help me with my essay."
What actually works: "Act as a Year 12 English teacher. Here's my introduction paragraph: [paste text]. Please: (1) identify whether my thesis is clear, (2) check if my hook engages the reader, and (3) suggest one specific way to improve flow. Explain your feedback at my level with examples."
The difference in output quality is dramatic.
AI literacy isn't optional anymore. It's fundamental. But we can't just give students access and hope they figure it out. They need explicit instruction in structured communication with AI systems.
Teaching this framework helps students:- Get useful feedback instead of generic responses
- Develop critical thinking about what makes requests clear
- Learn to be specific and intentional in their communication
- Understand AI as a tool that requires skill to use effectively
This isn't about replacing teacher feedback. It's about teaching students to use AI as a learning tool for brainstorming, getting unstuck, or checking their reasoning while understanding its limitations.
Start simple:
- Model the four-component framework with your class
- Have students practice revising vague prompts into structured ones
3. Compare outputs from weak vs. strong prompts
4. Discuss when AI helps learning and when it doesn't
The goal isn't to make students dependent on AI. The goal is to make them literate users who understand how to communicate clearly with any tool, human or machine.
As the creator of the first AI in Academic Writing course, I've learned this: students don't need less structure with AI.
Just the opposite, they need more. The four-component framework (Role → Input → Steps → Knowledge Level) gives them that structure.
Our job as educators is to teach students how to learn with AI, not just from it. That starts with teaching them how to ask better questions.
For educators implementing AI in writing instruction: I've developed templates and lesson plans based on this framework.
Lee, D., & Palmer, E. (2025). Prompt engineering in higher education: A systematic review to help inform curricula. International Journal of Educational Technology in Higher Education, 22(1), 7. https://doi.org/10.1186/s41239-025-00503-7
Walter, Y. (2024). Embracing the future of Artificial Intelligence in the classroom: The relevance of AI literacy, prompt engineering, and critical thinking in modern education. International Journal of Educational Technology in Higher Education, 21(15). https://doi.org/10.1186/s41239-024-00448-3
Wei, J., Wang, X., Schuurmans, D., Bosma, M., Ichter, B., Xia, F., Chi, E., Le, Q., & Zhou, D. (2022). Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems (NeurIPS 2022), 35, 24824-24837.
