How Hypothesis is Responding to ChatGPT

Hypothesis recently hosted a Liquid Margins episode on ChatGPT. Over one thousand people registered for the webinar and over five hundred attended synchronously. It’s clear that this new generative artificial intelligence (AI) tool has touched a specific nerve among educators within the broader hype and backlash that we’ve seen.

Some educators are concerned about the potential abuses of such technology by students. Others about the seeming obsolescence of the foundational skills many of them teach. ChatGPT produces passable prose on almost any topic in seconds! Some educators may see their own jobs at risk: ChatGPT can write essays but it can also write essay prompts and provide feedback to students on their own writing. Still, others are excited to explore the ways that tools like ChatGPT might expand and deepen how we teach and learn.

When new challenges arise in education, Hypothesis always turns to our users to inform our approach. We were heartened to see that OpenAI has similarly reached out to educators for feedback. The guests on our Liquid Margins episode–Joel Gladd from College of Western Idaho, Kat King from Diablo Valley College, Nick LoLordo from the University of Oklahoma, and Rachel Rigolino from SUNY New Paltz–shared a variety of pedagogical strategies educators could use to respond to the advent of everyday AI writing bots like ChatGPT and have been very influential on our thinking on this topic.

It’s hard to avoid concerns about plagiarism with the rise of ChatGPT. There’s no doubt the platform makes it very easy to generate text on a subject with only a simple prompt, like an essay topic. Plagiarism detection software like TurnItIn has promised to keep pace with Large Language Models (LLMs) like ChatGPT in what will become a war of algorithms. Hypothesis could integrate plagiarism detection software into our social annotation tool, but it’s likely a losing battle. From our perspective, plagiarism can be avoided and, more importantly, learning can flourish, when educators and technologists emphasize process over product.

In this moment of generative AI, Hypothesis continues to rely on what we’ve always done: support process-oriented pedagogies that make learning more accessible. Focusing on process over product also shifts discussions of academic integrity away from concerns about whether students are cheating toward a deeper understanding of how knowledge is produced and how students can be active participants in that process. What does it mean to read closely? What does it mean to write critically? Furthermore, we believe a process framework can help educators both work against the potential problems of ChatGPT and also possibly harness some of its affordances.

In terms of the written products of academic work, focusing on process over product means paying close attention to what John Warner has called, in this context and beyond, “the messy and fraught process of learning how to write”, rather than the final product of a five-paragraph essay. What if instead of a high-stakes writing assignment like an end-of-term essay, we introduced students to and guided them through the reality that writing is part of a longer journey composed of many low-stakes activities?

Building in more regular formative reading/thinking/writing activities creates more frequent touch between students, course content, their own ideation and writing about that content, and instructor (and possibly peer) feedback on their words and ideas. It shows them that writing, and learning more broadly, is a process rather than a product. Obviously, Hypothesis social annotation can help by making visible–both to instructors and students themselves–the ways in which students engage with readings for a class, their earliest thinking and writing on the course content.

It also seems possible that ChatGPT and other AI content generation and chat tools can help students with the learning and writing processes. It’s clear that the platform’s outputs can contain misinformation and bias, so should never be used as a final draft. But if we are thoughtful and deliberate in how we interact with the writing of ChatGPT–just as we encourage our students to read, think, and write critically about any text–we are not only helping them further their reading, thinking, and writing skills, but we’re training them to be responsible digital citizens and humanists. Again, we think social web annotation can help, offering students and teachers a practical means to engage critically with the text of ChatGPT as well as the debate around it.

We’ll elaborate on the specific ways that a process-oriented tool like Hypothesis social annotation can work with and against ChatGPT in a follow-up post, so stay tuned! In the meantime, if you are interested in connecting and conversing with other educators on the topic of ChatGPT and teaching and learning, check out this reading list by Anna Mills and annotate the readings with Hypothesis using the tag ChatGPTedu.

Check out our follow-up post Five Ways to Use Social Annotation With and Against ChatGPT.