Thinking with AI

How Inquiry-based Learning Informs Our Relationship with ChatGPT (and Software Like It)

by DeVaughn Harris

In the words of Albus Dumbledore, “Words are, in my not-so-humble opinion, our most inexhaustible source of magic.” What happens, though, when words are not solely produced by humans? Of course, I am referring to the proliferation of Artificial Intelligence technologies as they relate to the craft of composition—specifically, ChatGPT. The very act of writing has been appropriated from humans by these technologies, and the question that has reverberated throughout the academic community regarding that appropriation is: Where do we go from here?

I think we go back to the roots of composition by learning to become better questioners and thinkers when it comes to our writing because at a fundamental level, I truly believe that writing and imagination are interconnected. The ability to enter a mediative state, communing inward to find the right word, idea, and statement is central to the craft of writing. Lisa Blankenship, in her essay “Time-Saving Grading Practices,” refers to this type of learning that grounds itself first and foremost in the philosophically rich endeavor of asking questions about one’s own experience as “Inquiry-based learning.” This type of learning not only allows students to dig deeper into the choices they make as writers, but it also allows them to develop a deeper understanding of how they are making those choices, via the prompts that we impose upon them as instructors and the metacognitive prompts they impose on themselves as writers. When it comes to AI and composition, we must learn to identify and preserve the deeply human aspects of writing in the twenty-first century.

In 2023, the MLA-CCCC Joint Task Force on Writing, establishing an overview of the current state of writing instruction alongside AI technologies. In this paper, the authors state that one of the risks present for students in relation to AI is that “Students may not see writing or language study as valuable since machines can mimic these skills” (MLA-CCCC Joint Task Force on Writing and AI, 2023). Insofar as risks go, this one seems clear-cut. Imagining that students, who are often already overbooked in terms of course scheduling and workload, can triage their writing assignments by getting software such as ChatGPT, it isn’t far-fetched to think that they may begin to see less and less value in critically engaging with their writing selves and the writing process. However, there is a silver lining here: inquiry-based learning. Blankenship’s term exposes a crucial, human-specific point present in both analog and AI-assisted writing processes: the art of questioning. Even when students choose to use AI to assist, supplement, or generate their writing, they must start with a question or prompt. That moment is a critical point of entry. They begin with a moment of agency that, when fostered and encouraged, can inform how they engage with AI in what I understand to be a more authentic, human manner.

What makes better questioners better writers, and how can that relationship inform pedagogical innovations related to the use of AI technologies regarding writing? I turn to Peter Elbow’s musings on student writers’ ability to make choices to explore this question. In a response to David Bartholomae, Elbow invokes faith as a methodological launchpad from which we can begin to encourage students to make choices with their writing. He writes, “I must call on some faith in the ability of my students to make important choices, decisions and perceptions of their own when I can clear good space” (91). As educators, we should be extending a similar kind of faith to our students regarding AI and writing. That means, simply put, that instead of assuming that our students are using AI to subvert the writing process, and teaching from that perspective, we can have some faith that our students are asking questions about the writing process and their own writing before interacting with AI software. If our faith is positioned in such a way, we can then enact an inquiry-based learning approach toward using AI as a tool that can aid in the writing process.

Think of the writer’s cover letter. This letter is typically written after a writer has undergone peer review, workshop, and first-draft assessment—yet I believe that the pedagogical benefits of the cover letter can also be extended to an inquiry-based approach that precedes the crafting of an essay. Blankenship’s essay notes that the key idea with cover letters is “to guide and promote self-reflection and to help students understand the value of thinking critically about one’s rhetorical choices.” AI can serve in a similar capacity. I teach my students in ENGL 2100 to use ChatGPT as a tool rather than a crutch insofar as they should field any questions, they have about a writing assignment and/or their writing through the software. Some questions we have experimented with as a class are: “How can I refine my argument to make the point that I believe ‘X?’” “Can you explain this prompt for me?” “What are some ways to use Peter Elbow’s ‘believing game’ in a critical analysis?”  When my students have prompted AI in this way, they typically receive responses that help guide and promote serious reflection in their writing and essay prompts. This, in turn, encourages them to think deeply about their writing at the sentence and paragraph level. In other words, students are learning from AI to ask better questions of themselves and their writing.

My point is that an inquiry-based approach attends to one of the core tenets of Writing Studies, which assumes that better questioners and thinkers make for better writers. It is worth noting though, that my view is utopian insofar as it assumes that AI software is without fault. This is not the case. Software like ChatGPT is faulty—it can, for example, produce responses far afield from the entered prompt (which can be due to a number of factors, such as inherent coding bias related to things such as language difference). A reliable hot-fix to this faultiness, I think, lies in our ability, as educators, to teach our students to think deeper about the questions they want to ask. This can be enacted through mind-mapping sessions, during class discussion, wherein I ask students: What questions about the assignment do you want to ask ChatGPT, and why that question? This allows students to enter a metacognitive state in which they are aware of what they want to say, what they want to write, and how they imagine the verbiage playing out. Now, they just must figure out how to construct their imagination on paper—that’s where software like ChatGPT comes in. Some might say that this task is typically performed by the teacher, and that is fine. However, in the twenty-first century, in the age of mass writing and mass access, students have options. This does not mean the writing teacher becomes vestigial by any means; there is still a need for instruction, albeit supplemental, to help guide students toward ethical and useful ends. 

Below, I have included an example of a response from ChatGPT that follows from the prompted question: How can I make the point that Donald Glover’s ethos, on the album Camp, constructs an authority that is able to speak on cultural issues in the Black community more significant to an essay prompt that’s asking me to critically analyze Glover’s album through the lens of Peter Elbow’s “believing game?” ChatGPT’s response is as follows: