People utilize different metaphors to articulate their connection with AI. Some perceive AI as a largely dependable intern, while others regard it as a digital assistant. More frequently, chatbots like ChatGPT are evolving into friends, therapists, or even romantic interests. As a university writing instructor, I consider AI as a partner: a dynamic repository of information. Nonetheless, as a sober alcoholic, I also view it as a well-functioning drunk: occasionally sounding insightful without fully grasping the topic.
I have anecdotes about AI saving me time by managing tedious chores, proofreading, or engaging in discussions about my research interests. However, there are instances when it cheerfully misinforms or struggles to comprehend me, persisting in dialogue instead of acknowledging its mistakes. For example, I once requested ChatGPT to transform my academic conference comments into a presentation on literary journalism, and it generated slides on luxurious travel in Brazil.
Such occurrences serve as cautionary narratives for my students. While I feel AI can undermine significant human motivations for writing, not every piece of writing is identical. Writing frequently entails research and necessitates critical feedback post-drafting. Rather than adopting a reactive approach, I aim to investigate with my students how AI can function as a productive collaborator in this endeavor.
Engaging with the archive
University writing strongly depends on research and reading, training the mind to organize data and think systematically. Employing contemporary technologies for this purpose doesn’t negate critical analysis. Throughout my life, these advancements have transitioned from library card catalogs and microfiche to online resources like JSTOR and Google Scholar. These tools don’t diminish thought; they enhance brainstorming and data collection, broadening the information we can access.
Having observed the swift digitization of research and writing tools prior to AI, I tend to regard AI as a collaborative research ally. In my domain, literary scholars invest considerable hours in libraries and archives. Digitization has enhanced accessibility to these resources, and AI may facilitate their analysis.
I’ve begun perceiving conversations with an AI chatbot as interactions with an archive. Before engaging in intensive work, we can participate in a research-oriented dialogue with a “mind” that holds a general understanding of available data. Recently, I employed ChatGPT’s advanced voice feature to inquire if likening chatting with the archive to conversing with AI was an apt analogy. It responded, “When you’re engaging with an AI like me, you’re tapping into a vast reservoir of information and patterns derived from human knowledge up to a certain point.” It also emphasized, “It’s crucial to remember that while I can deliver information and insights based on that knowledge, I lack human experiences or consciousness. So, while it may seem like discussing a vast well of knowledge, it’s always wise to consider the human perspective and context as well.”
As our dialogue deepened and my inquiries became more targeted, I could request references and additional reading ideas. This pre-writing exchange with AI is becoming integral to my workflow. I’ve consistently found it easier to cultivate ideas through discussion, but few individuals show interest in my half-formed concepts. That’s why brainstorming is one of the most valuable applications of AI for authors.
Establishing your own mini-archive
Interacting with AI has proven beneficial for generating ideas, and the transcript makes revisiting them simple. There are also AI-oriented tools designed for more rigorous research tasks. At the end of the fall semester, a student reached out to me about Google’s NotebookLM. I wasn’t familiar with it, but I quickly grasped the concept. NotebookLM extends the idea of engaging with the archive: you build your own archive for a particular project, and the AI can assist in gathering sources to commence.
For my recent conference presentation, I uploaded 25 PDFs saved in Zotero, my preferred citation manager, into NotebookLM. It swiftly “read” them and provided a summary: “These sources discuss ordinary language philosophy, primarily focusing on Wittgenstein, Austin, and Cavell, and its relation to other philosophical and literary movements like pragmatism, transcendentalism, and deconstruction.” Below the summary is a text input field inviting me to “Start typing…” with suggested prompts such as, “How does ordinary language philosophy challenge traditional philosophical approaches to meaning?”
On the right side of the page, in the “Studio” section, I’m encouraged to create an audio overview, similar to a podcast featuring two voices discussing my topic. In Interactive Mode, I’m treated like a caller on a late-night radio show, receiving praise for my questions and responses based on the documents I provided. The podcast functionality isn’t superb yet; it’s somewhat flattering, but it has the potential to improve and become more effective. NotebookLM offers additional features: generating a “Mind Map,” study guide, briefing document, FAQ, and timeline. I will continue to utilize it and recommend that students do likewise.
A not-quite writing tutor
AI could assist student writers by offering immediate feedback on their writing. When I inquired about this with ChatGPT, it proposed considering AI as a “writing tutor that’s available 24/7,” with the caveat that it “lacks the