Teaching Writing with Generative AI

purple abstract art

This post is based on a presentation I ran recently at the AATE conference. It takes the writing cycle approach from our book ‘Practical Writing Strategies‘ and applies Generative AI at various stages.

One of the most pressing concerns at the recent Australian Association for the Teaching of English (AATE) conference was the implications of Generative AI for teachers. There were multiple panels and sessions on the topic, ranging from discussions of AI Ethics to practical considerations for English teachers working with (and sometimes against) GAI technologies like ChatGPT.

I ran two sessions, one on writing instruction, and one on Generative AI. The two were closely related, and both based around the writing cycle from the book I wrote with Ben White, Practical Writing Strategies. In the book, we explore ways that writing can be seen as both a process and a product, and we use the following writing cycle as a loose structure to guide students through their writing:

The PWS Writing Cycle

I’ve explained the writing cycle elsewhere so I won’t go into it in much detail here, but it has proven to be an excellent way to encourage students to take the time and effort needed to produce quality writing whether that’s fiction or nonfiction, print or digital.

GAI and the Writing Cycle

In the AATE session, I outlined ways that Generative AI can be used across the six different stages to support, but not necessarily replace, student writing. Combined with the AI Assessment Scale which clearly articulates how students can use AI in a given task, it offers a great way to guide students through the appropriate use of AI in their writing.

For this session, I used ChatGPT (3.5), but you could easily try this with Microsoft’s Bing, Claude’s Anthropic, or Google’s Bard with similar results. The point is to allow students to use GAI as a tool: sometimes offering up suggestions or brainstorming ideas, sometimes assisting with editing and proofing, but never replacing their own personal voice.


In the initial stage of writing, we clearly articulate the purpose, audience, and context for the text we’re producing. For students, the purpose to go beyond “because it’s an assessment task”. If we can construct authentic audiences it makes it much easier to conceptualise the purpose of the audience. Prompts like these, for example, could be used with GAI to create audience personas to write for:

Identify five possible audiences for a blog post about the climate crisis.

Role play: You are a member of the audience for my blog post about the climate crisis. You are a <insert details about audience member here>. Give me a detailed audience profile and tell me your three burning questions about the issue you’d like to see me address in my writing.

Here’s how that might look:

Extend: Extend this discussion by challenging the stereotypical or potentially biased output of the model. Why is the environmental activist a 25 year old woman named Sarah? This could easily turn into a great discussion of target audiences, demographics, and stereotyping in the media as well as highlighting the tendency towards bias in AI language models.


In the exploration stage, students discuss and examine mentor texts similar to the writing they’ll ultimately create. The point at this stage is to get a feel for the style, voice and techniques. You might use various close reading activities to explore these texts. GAI can be used both in the creation and analysis of mentor texts.

How does this author use descriptive and figurative language? List 3-5 features of the text: <copy/paste text extract> (Respect copyright!)

Take point 2 and expand it into an exercise I can use to practise that skill.

Extend: In the above example, the model is being used to interpret a mentor text the student has provided. By flipping the prompt around and using those elements, for example “use sensory details, figurative language, vivid imagery…” the model can generate novel mentor texts to explore.


In the ideas stage, students brainstorm their own writing in response to the purpose, and based on the style and techniques explored through the mentor texts. GAI can assist with the brainstorming process, particularly if you’re short on class time and just need to focus on getting students writing.

Here’s my basic idea for a story about <topic>: <idea>. Brainstorm 10 possible sources of conflict in this story.

Generate 5 ideas for how a character might tackle the fourth conflict.

Extend: Consider how a Thinking Routine like Connect Extend Challenge might fit into this process. A student could generate ideas by hand, and then use the GAI to help sort and strengthen their ideas with the Thinking Routine.


Skills development can happen throughout the writing process: the act of writing itself is one of the best ways to sharpen the tools. But you can also dedicate time to teaching explicit skills such as sentence structure, functional grammar, and vocabulary. GAI is useful in this scenario for students’ self-assessment and self-directed skills development.

I need to work on my sentence structure. What advice would you give to help me improve the following? <copy/paste draft writing>

I’ve worked on that. Here is the second draft: <copy/paste draft>. Please provide more feedback and suggest other areas to work on.

Extend: Use this as part of a sequence of conferencing lessons where you work closely with individuals or small groups, and the rest of the class uses GAI to focus on their own skills and needs.

Success! You're on the list.


Not many texts are written in isolation (except senior secondary school exams…). When I write a book, it goes through an editing process that involves at least one and often two or three others reviewing the work. Even when I write blog posts, it’s in concert with the audience I imagine reading the finished piece. GAI can be used to add an extra dimension to the collaborative workshop and peer assessment process.

Based on the following criteria, create step by step instructions for a group writing workshop: <copy/paste criteria>

One of our group members is absent. Fill in for them by helping with the peer review section. Please provide constructive feedback on this: <copy/paste draft work>

Extend: After using ChatGPT as an extra set of eyes on their work, open up the critical conversations again. What did ChatGPT miss? Where did it focus the attention? Did it display any overt or subtle biases in its response? When is human feedback better?


Outside of education, texts genuinely have a real purpose. Inside of education, we might have to manufacture that to go beyond “because it’s the assessment”. But it can be done. Writing competitions, school anthologies, literature class critical journals, poetry slams, school newsletters, blogs and other forms all offer students opportunities to publish their work. GAI can help prepare students for publication.

I’m a secondary school student in Australia. We’ve just completed a writing assessment on <topic>. Suggest 10 places I could try to get my work published.

Guide me through the process of polishing my work for number 3.

Extend: Don’t stop at just teaching the process of publication: as often as possible, try to find ways to get students’ writing out into the wild.

Pulling it all together: an example unit and assessment task

These technologies can seem very abstract if you’re unfamiliar with them, so I’m going to contextualise the writing cycle and GAI through an example unit. This is a six week analytical writing unit, the outcome of which is a “literary critique” style piece of analysis that could be published in a class journal. Students could write about any text they are familiar with, e.g., a text from a previous unit of work, or something they personally enjoy.

WeekContent and questionsGAI use
Week 1Purpose: What are we writing? Why are we writing it? Who is it for? What is the purpose of analytical writing and literary critique?Generate a list of conventions for literary journals. Generate “literary critic” personas to explore.
Week 2Exploration: What does analytical writing look like? What are the conventions?Generate short analytical mentor texts based on texts students are familiar with. Aim for a variety of voices and styles.
Week 3Ideas: What will you write about?Assistance with brainstorming and organising ideas.
Week 4Skills: What do you need to know how to do? What do you need to work on the most?Self-assessment to identify skills gaps. Exercises based on specific skills, e.g., writing in a formal register or using nominalisation and the passive voice in objective writing.
Week 5Collaboration: How can you help another writer? What do people think of your work?Scaffolding for peer assessment and workshops. Generic feedback and editing (like Grammarly).
Week 6Publication: Is it ready to publish?Final proofing and advice.

Amongst all of the possible GAI use in that unit the students’ own writing should remain at the centre. You’ll notice that none of these prompts include anything like “write me a draft based on these notes” or “answer the following question on this text”. While it’s entirely possible for students to use the technology like that, we should be aiming for engaging writing units where they want to do most of the thinking and writing for themselves.

Hopefully this post serves as one example of a way GAI can be embedded into a writing process without entirely replacing the student’s need to write. I’m sure there are many more ways we can appropriately and ethically bring these technologies into the classroom.

If you have comments or questions about this post, or would like to discuss working together at your school, please use the form below:

Leave a Reply

%d bloggers like this: