In the Deep End of Data: How is ChatGPT Catalyzing Change in Education?

Beth Dewhurst

Paul Magnuson

ChatGPT is one of the most exciting innovations in education we can remember. And overwhelming, and stress-inducing一mostly because its uses, challenges, and implications range far beyond the scope of our current understanding. We certainly cannot keep up with all the posts about ChatGPT these last few weeks, and we doubt other bloggers and writers can either. Case in point: Jason Murphy explained on the weekend how he was experimenting with ChatGPT for teacher comments and grading personalized to individual learners. By Wednesday of the following week he wrote on LinkedIn: “Ignore my posts. Read this instead.” An EdTech developer in South Korea had posted a manner in which teachers can integrate their LMS and ChatGPT to handle many student comments, quickly.?

Educators are testing ChatGPT for more than creating student feedback. The chatbot can create lesson plans, suggest ideas, make test prompts, develop essay questions, and far more, all in seconds (with intermittent bouts of wait time for user traffic to clear). We are pleased that the conversation is stretching far beyond surveillance (a Sisyphean cycle of countermeasures and AI evolution) to focus on how ChatGPT can spark new ways of thinking, teaching, and learning. If we are serious about teaching 21st century skills, after all, we need to incorporate 21st century technologies.?

If it all feels like too much to wrap your head around, that’s probably because it is一especially if you’re trying to work out the implications without an exchange of ideas among colleagues, your learners, and others interested in asking a lot of what-ifs. The two of us certainly can't shake the feeling that we understand so little about the complex, evolving implications that ChatGPT represents for our teaching practice. We are going to need to carve out dedicated time to connect with each other, specifically about the use of AI, to help process what we read and what we hear. Courses are likely to spring up, like AI, Chatbots, and ChatGPT for Teachers, developed by the International School of Geneva, and over time, guidance for reasonable ways to leverage ChatGPT will get more specialized. While we haven’t seen a lot of discussion specifically about chatbots and online learning environments, we know it’s only a matter of time, measured in days, not weeks.?

Accessible, doable ideas will keep surfacing, like those of Rowena Harper, a researcher/practitioner in academic integrity and literacies. “Let's interrogate when, why and how we ask students to reproduce existing knowledge: what must they know to enact higher-order thinking, and what can remain at their fingertips.” (Harper, January 13, 2023). In mid December, Ryan Watkins, Professor of Educational Technology Leadership, and Human-Technology Collaboration at George Washington University in Washington, DC, advocated that prior to revising or reimagining learning activities or assessments, our first step is to clarify our objectives before infusing them with creative updates using the bot. By adapting our practice and incorporating ChatGPT and other AIs, we facilitate the possibility with our learners about how we, collaboratively and in real time, can privilege learning through the use of the chatbot.

Watkins suggests, for example,? a dual writing assignment in which learners can opt to use ChatGPT or not. Those who choose to use the chatbot revise the text with a track changes tool to indicate how they improved the bot’s output. Those who choose not to use the chatbot sign a statement that their learning products are bot-free. Both can be given feedback on the qualities of their responses. Watkins points out, “ChatGPT is just the most recent AI system to be released, there are many more in development and they will have varied implications to your teaching in the future. Scan the Internet, talk with colleagues, and chat with students about the newest technologies.”?

Teachers are also noticing the bot’s limitations. Prompts that include too many specifics produce increasingly generic results (though there are a myriad of sites and even a prompt management system underway to help cull and organize the most effective prompts).? Additionally, the degree to which the bot can recognize context or yield some forms of creativity may currently foil ChatGPT. As San Francisco State University’s English Language and Literature lecturer and poet Brian Strang explained, “Currently, this bot seems like more of a threat to search engines than to writing teachers. If a student tried to pass off this kind of writing as their own, they probably wouldn’t get far in the writing classes we teach . . . I want students to dare to have an opinion, develop their own unique voice, express themselves and support their arguments in ways that are effective and which fit the particular audience, purpose and context within which they write at any given time.”?

In her streamlined collection of excellent articles, ChatGPT Advice Academics Can Use Now, Susan D’Augustino concludes “Human learning is gradual, even when AI learning seems instantaneous. That will not change, so teachers will likely be the most important users of AI writing tools. We will mediate, introduce and teach. So, our advice for colleagues is straightforward: start experimenting and thinking now.”? We have an opportunity to create new ways of building knowledge with our colleagues and learners if we’re willing to collaboratively build the systems for interpreting data and making decisions that our AI learning ecosystems require.