This is the fifth post in a series exploring how Generative AI can be incorporated into the writing cycle from our book Practical Writing Strategies. Check out the previous stages of the writing cycle: Purpose, Exploration, Ideas, and Skills.
Collaboration in the Writing Cycle
By the time students reach this stage in the writing cycle, they will no doubt have collaborated on various aspects. For example: brainstorming ideas in groups, annotating mentor texts with a partner, and classroom discussions of purpose, audience, context, and the skills needed to improve their writing.
In stage five, there is a deliberate focus on collaborating with each other and with generative AI. At this point, students should have identified the purpose and audience of their writing, decided on an appropriate form and style, and have begun to write draft material and hone their craft through skills development. The collaboration stage, then, is focused on editing, feedback, and revisions
AI-Assisted Editing
In the artificial intelligence assessment scale developed in collaboration with Mike Perkins, Jasper Roe, and Jason MacVaugh (currently being trialed at British University Vietnam), we include AI-assisted editing on level three. This means if any assessments place at this level, students are free to use artificial intelligence tools to edit and refine their work. I believe that this is going to be one of the most common uses of generative AI for the near future.

Because of the huge corpus of language data, large language models (LLMs) like GPT-4 and Gemini are more than competent at reproducing Standard English and a variety of languages. While they struggle with dialects and languages that occur less frequently in the dataset, their ability to produce competent texts makes them great tools for editing.
However, there is an important caveat. When using these tools for editorial purposes, they have a tendency to overwrite rather than correct. We don’t want to see Standard English at the expense of the student’s voice. I wrote about this in my article on the “myth of the AI first draft”, including the risk that students might see generative AI output as inherently better or more correct than their own.
When we’re using these technologies, we need to stress to students that their voice is still of paramount importance. The voice of the AI model is a disembodied machine representation of language based on probability and with no connection to the real world, to emotion, to memories, or to personal experience. Whether students feel confident in their writing or not, whatever they write is just as important, if not more so, than machine-generated content.

Practical AI Strategies is available now from Amba Press and bookstores online
Activities
Activity 1: AI Editor
This is a straightforward use of language model applications like ChatGPT, Microsoft Copilot and Google Gemini to catch simple errors, such as spelling, punctuation, and grammar. Importantly, the prompt includes a specific instruction not to rewrite the original text.
- Step one: Discuss with students the importance of their voice as outlined above. And the fact that whilst AI generated content might be grammatically correct, it is not inherently better.
- Step two: Students take an extract of draft writing and pass it to an AI application such as Copilot or Gemini with the prompt below.
- Step three: Students review and respond to the feedback, making edits if they feel they are appropriate and ignoring them if not.
review this piece of writing and make a dot point list of suggestions focusing on spelling, punctuation, grammar and syntax. Do not rewrite or correct the passage and do not make suggestions beyond these requests: <copy/paste text>

Activity 2: Peer Review
In academic writing the peer review process is designed to bring an impartial eye to work to ensure its credibility, the strength of the arguments and the clarity and logic of the writing. For students writing nonfiction, including argumentative pieces, a genuine, anonymous peer review can offer powerful feedback.
For the classroom teacher this is difficult to orchestrate: even if students leave their names off a piece of work a teacher can often tell who has written the text thanks thanks to familiarity with their writing style (or with their handwriting if it’s written on paper). This activity avoids the problem by using generative artificial intelligence. GenAI knows nothing at all about the writer (in fact, GenAI knows nothing full stop…). It’s also possible to generate multiple peer reviews ant once, including the dreaded “reviewer 2”.
- Step one: This exercise requires students complete a full draft first. It works for both fiction and nonfiction, but the feedback is particularly useful for nonfiction and argumentative writing.
- Step two: Submit the draft to a large language model based GenAI with a prompt like the one below.
- Step three: Respond to feedback as desired.
You are a peer reviewer: your role is to review the following text particularly focusing on clarity, logic, the strengths of the arguments, and connections between ideas. Once you have reviewed the text in full, change roles and become another reviewer to provide a different perspective. Both views should include a mix of positive feedback and critique. <copy/paste draft>

Activity 3: AI Image Editing
This activity serves a dual purpose: firstly, it teaches students how to use digital editing tools to alter their own images. Secondly, it alerts them to the fact that anyone with access to these tools can alter images in this way, adding an element of critical media literacy. This activity is suitable for any text which includes images and other visual elements.
It requires a combination of tools including access to a large language model based application like Microsoft Copilot or Google Gemini which has image recognition features, and access to Adobe tools such as Adobe Firefly, or Adobe Photoshop. All of the Adobe Creative Cloud tools are accessible for students and educators in K-12 and tertiary on a single institutional education license.
- Step one: Students create images in a platform such as Firefly or Copilot.
- Step two: Using an application with image recognition (Copilot, Gemini, ChatGPT Plus), upload the image and ask for suggestions on how to make it more compelling, more persuasive or more suited to a particular theme or idea. The language model will offer suggestions which the student can then use to edit the image.
- Step three: The student can now use GenAI in a variety of ways:
- Microsoft Copilot: upload the original image and the feedback and ask the model to generate a new image based on the original and incorporating feedback. The results from this will vary. Sometimes the model will generate an image which is very similar to the original with only slight tweaks, and other times it will generate a totally new image. It is worth trying a few generations to find a suitable image.
- Adobe Firefly: upload the image as a “reference image” and write a new prompt based on the feedback.
- Adobe Firefly or Adobe Photoshop: open the original image and use the generative fill or generative expand tools to edit the image directly. This is the most advanced option and will require some time to get familiarity with the tools.

Activity 4: Writers’ Workshops + AI
This activity is adapted from Practical Writing Strategies. The writer’s workshop (+ AI) is a collaborative activity where students can share their writing projects and receive feedback from their peers. The goal of the workshop is to provide a supportive and collaborative space where students can develop their writing skills and improve their projects. AI is used in an administrative capacity to keep the conversation between students as the priority.
- Step one: Begin by introducing the concept of the writer’s workshop and explaining its purpose. Emphasise that the goal of the workshop is to provide a supportive and collaborative environment where students can share their work and get peer feedback.
- Step two: Divide the class into small groups of three to four students each. Use a device with audio recording capability: this can be a phone with a voice memo app, a laptop with a microphone, or any other device.
- Step three: Ask each group to choose a writing project they are currently working on (narrative, science report, persuasive speech). Begin recording (ensure there is permission to record, and ask students not to use full names – always avoid personal identifying information when using AI tools). Each student takes turns sharing their writing with the group. As they share, the other members of the group listen actively.
- Step four: The listening students should discuss two strengths they identified, and suggest two improvements.
- Step five: After each group member has finished reading, the recording can be processed and transcribed by AI. I use otter.ai which has both free and paid licenses. There are many other AI transcription tools available.
- Step six: The “raw” transcript might be hard to follow, so it may be useful to process it further by passing it to a GenAI platform with a prompt like the one below:
Format this transcript. Make the text clear, correct any obvious transcription errors. The most important features are the strengths and suggestions for feedback: prioritise these aspects of the transcript in a clear list.

Conclusion
These activities are designed to facilitate collaboration with an AI model or encourage collaboration amongst students and emphasise the importance of the editing stage of the writing process. Students often rush to hand in work, and unfortunately, this is a byproduct of the system that we’ve created around writing assessments. We can change this by valuing all the stages of the writing cycle.
The next article will be the final in this series, focusing on publication and looking at how generative artificial intelligence can be used to make writing tasks applicable to the real world, authentic, and valued by students.
If you have questions or comments, or you’d like to get in touch to discuss GenAI consulting and professional development, use the form below:

Leave a Reply