Leon Furze

With Victoria the latest state to ban ChatGPT, teachers everywhere are grappling with the implications of the large language model and the impact it will have on the classroom.

We know that banning and blocking technology is – at best – a temporary solution. Students will find ways around the blocks, and teachers will be faced with policing the technology and its use. Aside from the “cheating narrative”, many teachers are also very hopeful of how the technology could be integrated into their classrooms.

Last week I put out a survey to teachers asking about their concerns and hopes for the future. Here’s some of the feedback:

• Teachers believe that if students are taught about ethical and moral obligations and how to use AI appropriately, it could change the way writing is taught in the classroom.
• Some teachers are concerned about the loss of motivation in students and the idea that “when a bot can do the job in 20 seconds”, students will lack the intrinsic desire to write
• Other teachers want to learn how to integrate AI into the classroom, working with it to help students refine and edit their work instead of banning it.
• Some teachers are interested in using AI to assist students in their learning and to assist teachers with planning.
• Finally, teachers want to learn how to show their students that AI is a tool and not the whole process, and how to find new ways to assess their writing skills.

This post starts to address some of these hopes and concerns.

Academic integrity and living with AI

If we take banning and blocking off the table, we need to be prepared for some honest and open conversations with students about academic integrity. Essentially, this is nothing new. We have been speaking to our students about copying, plagiarism, and getting other people to do their work for them for decades.

I’m a PhD student at Deakin University, and last week an email landed in my inbox with this simple, clear advice:

If you want to explore using ChatGPT or other AI tools to assist with your assignments or research, you should:

1. Use AI as a tool to assist you in your research and writing, but not as a replacement for your critical thinking and analysis.
2. Ensure that you appropriately cite and reference any text or output generated by AI in your assignment, along with any other sources you use. You should clearly indicate where in your assessment task you have used AI-generated material.
3. Understand the AI tool’s limitations and use it in conjunction with other sources to ensure the credibility and reliability of the information you present. You need to check the accuracy of all information generated by AI tools.
4. Be aware of the University’s student academic integrity policy and ensure that you follow it.
5. Make sure that the final product is your own work, and not just copied from an AI generator. You can use the generated text as a prompt for inspiration or guidance, but the final submitted assessment must be your work, creation, and analysis.

Deakin University advice on using AI like ChatGPT in assignments

I think this makes for a great starting point in updating your own school’s academic integrity policies. It acknowledges the benefits and the limitations of the technology, and suggests clear and appropriate uses.

Once these guidelines are established, I think the conversation around writing assessment forks into two directions: writing that AI can’t do, and writing that students should be doing with AI.

Join the mailing list for more posts like this, and to stay up to date on upcoming PD:

Writing that can’t be done by AI

Understanding the limitations of a model like ChatGPT is important for both students and teachers. Although the technology is impressive, and improving all the time, it will never be able to replicate certain human writing. Here are a few suggestions for writing assignments that it would be difficult or impossible for an AI to complete without sophisticated and dedicated prompting (which I’ll discuss later):

• In-class writing: though I’m no fan of “reverting to pen and pencil exams”, having students write in class, by hand, is a sure-fire way to avoid AI input. Rather than using this as a policing tactic, I think there are perfectly valid uses for in-class writing. Free writing; short, reflective writing like journalling; brainstorming; note-taking; and early drafts all benefit from the messy and human approach of writing by hand.
• Writing based on personal experience: As Nick Cave pointed out in his scathing review of ChatGPT’s artistic talent, an AI doesn’t have any lived experience. Our students do, however, and so a shift towards more personal writing guarantees a more human output. Have students reflect critically on the issues that matter to them ad their communities, drawing on lived experience, relationships, and memories.
• Reflecting on current affairs: At the time of writing, ChatGPT doesn’t have a live connection to the internet. Some GPT-based apps do, but for the most part it’s currently hard to get an AI to write about something which only just happened. Having students write and reflect on an issue which happened just this morning in your local community could be a great way to bypass the AI.
• Alternatives to writing: Writing has been placed on a pedestal as one of the highest forms of assessment, but AI has kicked it off its perch. Assess students’ knowledge in ways that can’t be easily replicated by a language model, such as orally through debate and discussion, using graphic organisers and visuals, and in collaboration with other students.

Writing that can be done with AI

For those teachers lucky enough to have escaped the blocking of ChatGPT, there are many exciting possibilities for working with the AI rather than against it. Here are a few suggestions:

• Editing coach: We often get students to peer assess each other’s writing, but neglect to teach students the real editorial skills required. Using a role play prompt two students could use an AI model as an editorial “side kick” to help them learn the skills of editing a text.
• Critical analysis: AI output is notoriously biased, but even when it isn’t overtly discriminatory it can be harmful. Have students analyse the output of an AI writer to look for potential “hidden biases” and assumptions. Provide the AI with prompts designed to expose these hidden biases and have students discuss what could be done differently.
• Idea development: AI models are great at outlining and developing ideas. Students could bounce ideas back and forth with an AI model while in the early drafting stages, and then move to their own writing before returning to the AI for editorial help.
• Human-in-the-loop: We know that fact checking is an important part of media literacy. This is even more true now that AI can produce text at such as high speed and volume. Discuss the importance of having a “human-in-the-loop” when reading materials online, and how humans can improve AI written text through editing and fact checking.
• AI comparison: Have students write their own text (including under in-class writing conditions) and then generate the output from an AI for comparison. What was the same? What was different? Did the AI output raise any critical questions about key perspectives or ideas that the machine missed?

Conclusion

Teaching writing in the age of AI offers many of opportunities for educators to integrate AI into the classroom in meaningful and effective ways. From exploring the biases and assumptions present in AI output, to using AI as an editing tool, or comparing and contrasting AI and human writing, there are plenty of possibilities for engaging students in a variety of learning experiences.

As AI continues to become more prevalent in our lives and in the way we work and communicate, it is vital that students learn how to critically evaluate and use AI in their writing practices. Teachers can help equip students with the skills they need to be informed, responsible, and effective writers in the age of AI.

Got something to say about this post, or a question about teaching writing int he age of AI? Get in touch: