Fooling Audiences with ChatGPT
I’m already sick of reading articles about ChatGPT. I know, that’s a bit hypocritical for someone who’s been blogging about ChatGPT, but give me a second to explain.
When ChatGPT blew up over this last month, the media hype-cycle was fairly predictable: news outlets made special video segments, tech reporters wrote think pieces, and clickbaiters barraged us with “The 5 ChatGPT Productivity Hacks You Need to Know!”
Amidst this ChatGPT media cycle, you may have noticed a frequently employed rhetorical tactic that, with some variations, went something like this: the article would open with a fairly straightforward description of ChatGPT or AI technology, and then the writer would break the fourth wall to reveal that the previous paragraph(s) was actually composed by ChatGPT itself.
Writers at The Atlantic are particularly fond of this tactic, where it can be seen in this article discussing the effects of ChatGPT on white-collar work, this one about the death of the college essay, and this article from Ian Bogost about ChatGPT not being as smart as we think. Bogost, however, is at least aware of the wider genre ecology of ChatGPT rhetoric he is participating within, writing that “...pretending to fool the reader by passing off an AI copy as one’s own, like I did above, has become a tired trope.”
In case you’re wondering, no I did not have ChatGPT write the introduction to this blog post...or did I? (No, I didn’t.) But I think what this trend reveals is that for any rhetorical endeavor, and perhaps writing in particular, it’s always a fine line between creativity and cliché. And to be clear, I’m not trying to attack any writers of these articles or their decisions to open their text with this “tired trope.” Rather, I simply want to consider how our critiques of Chat-GPT as a potential plagiarism machine reveal a fundamental misunderstanding of how writers develop a sense of genre, form, and structure, which do not emerge from the genius of a siloed mind but are absorbed through a thousand socio-rhetorical interactions (much like a Large Language Model) that came before them. In other words, much like a writer opening their article with the same trite introduction as countless others, ChatGPT is simply another reminder that the kinds of things we might think are unique or original about our compositions are perhaps more commonplace than we realize.
Assistive-AI will certainly continue to challenge our assumptions about the relationship between genre, style, and creativity, but at the same time, we also need to be reminded of Carolyn Miller’s assertion that writers are always participating in typified rhetorical actions (like writing articles, blogs, and videos about ChatGPT). In such cases, the reuse or adaptation of prior textual material is not a feature to be flagged and punished but perhaps an indication of a writer trying to find a rhetorical foothold within a complex kairotic moment.