Five ways to use Social Annotation with and against ChatGPT

Social annotation is an excellent way to make the reading, thinking, and writing processes of academic work more tangible and accessible to both students themselves and their instructors. As such, it can be a great way to both work against the potential abuses of AI chatbots and to engage with the written products of large language models like ChatGPT.

Here are five examples of using social annotation within the context of the rise of AI writing bots like ChatGPT.

1. Writing throughout a course rather than just at the end

If students are annotating throughout a course, across multiple readings, they are writing a lot! Their writing is part of the formative process of developing ideas, language, and arguments for a final paper or other summative assessment. Social annotation can be used to both scaffold toward and account for such summative assessments. Rather than being faced with an intimidating blinking cursor on a blank page, students have their own material to draw from for a final paper.

Furthermore, when teachers engage with student thinking and writing throughout a course, instead of only at the end, they can provide more frequent guidance in the process and get to know their students more individually. While ChatGPT offers impressive affordances in terms of modulating the style of its written outputs, it cannot imitate an individual student’s voice. We should be familiarizing ourselves with, and nurturing, our student’s writing styles and lines of inquiry. Acknowledgement and encouragement of one’s own ideas and words is part of the pleasure of and motivation for writing!

2. Writing in context, writing in community

Annotation is writing in very deep and specific contexts within broader texts: specific passages, specific phrases, specific words, even specific letters. Such specificity makes it less easily hackable for ChatGPT. Interestingly, one of ChatGPT’s shortcomings is its inability to integrate quotations effectively, and it cannot have a conversation with another author or text. The same specificity of reading and writing in context is also highly instructive for students learning to develop their skills of arguing from evidence.

That social context of collaborative annotation also makes it difficult for AI writing bots to reproduce. Students annotating a text with classmates have to be responsive to both the writing of the underlying author and their fellow readers. Perhaps more importantly, reading, thinking, and writing in community may better motivate students to read, think, and write for themselves. Rather than writing for a teacher for a grade, they are writing with and for their classroom peers.

3. Students annotating their own writing

Part of the problem that ChatGPT presents in education is in how teachers assess student work, though these challenges were not created by generative AI. A summative assessment causes both student and teacher to focus all their attention into a singular product that is supposed to stand in for a larger set of critical processes enacted over time. But with annotation that final product can also be a record of all those earlier stages in the act of production. When students annotate their own writing they gain agency in the assessment process rather than being merely a subject of it.

When students are asked to annotate their own writing, especially an essay that is the final product of multiple drafts and multiple sources of feedback (instructor and/or peer), they take ownership of their own writing processes. They have to re-engage with their own writing and explain their writerly decisions in ways that would be difficult if it was someone–or some “thing”–else’s writing. This type of metacognitive engagement with the process of knowledge production cannot be reproduced by an AI chatbot, though it could perhaps be applied to the writing of a tool like ChatGPT.

4. Students writing with AI writing tools

As others have argued, rather than banning ChatGPT or working to avoid or detect its usage, instructors should consider deliberately leveraging its affordances. Just like other tools used in the classroom, from spell check to calculators, there is obvious potential for abuse, but also a great potential benefit. It should be part of the digital literacy we teach our students to be thoughtful and ethical consumers of technologies like AI.

Perhaps ChatGPT can be used as a writing tool, to help further the processes of thinking and writing rather than somehow replacing them. Students could be asked to engage in a conversation with ChatGPT as part of their preparation for an assignment, perhaps even citing AI outputs within papers when appropriate. If instructors are concerned about the line between ChatGPT and a student’s original thought, they could ask students to annotate the transcript of their conversation and note how they were helped by the AI and what they learned in the process.

5. Annotating the writing of AI chatbots

Annotation can be used to fact-check, comment on, and critique the products of AI writing tools. The ShareGPT browser extension lets users make a stable URL of a ChatGPT session that can be annotated with Hypothesis. Let’s teach students to be critical readers (and perhaps editors) of ChatGPT outputs. What mistakes (or “hallucinations”) does the bot make? What does it get right? Does it show any biases? How is its writing formulaic? How can students improve upon what they are given by AI? If AI lacks an understanding of what it’s writing, can students add the meaning?

We are likely to see AI become more and more a part of everyday life, not only in education but in a range of professional workplaces. Helping students to understand how ChatGPT works, both its benefits and its limitations, and training them to work with AI tools, to train AI tools, will be a service to them both academically and professionally. For example, teaching students to write effective prompts for generative AI to respond to could be a very valuable lesson in AI-era education. In fact, if we look at the rubrics for effective prompts, they’re not too far off from the expectations of a conventional writing assignment: concision, clarity, detail, and evidence.

We would love to hear more about how you think Hypothesis can be used with or in place of tools like ChatGPT. If you have specific assignment ideas, please submit them to our new Resource collection.

To read our statement on generative AI in education visit our previous post How Hypothesis is Responding to ChatGPT.