Recently, I saw a presentation on the inter-Google that piqued my interest in AI. The pitch involved using AI to write fast, cheap novels you can hawk for sale on Amazon. Romance? Done. Mystery. Complete. Sci-fi? To quote the Terminator, “No problemo.”
As Zoey’s tutor, I’ve been closely following efforts to ensure academic integrity in the classroom since the introduction and rapid spread of AI tools. This is a hot-button topic in education. I’ve read that some teachers have even begun requiring their students to write during class to guarantee that the written product is authentic. Then there are programs like Grammarly, a utilitarian grammar checker I’ve used for several years, which now offers an AI plagiarism feature. It is ironic that this feature uses artificial intelligence to catch the cheaters. However, for students accused of plagiarizing, the consequences can be dire.
I decided I’d put Grammarly to the test with the following passage from one of my past blog entries:
“A recent New York Times article, ‘Two Drugs Stir Hope for Treatment of Deadly Pancreatic Cancer,’ opines, historically, ‘when it comes to pancreatic cancer, there are few treatment options, and those that are available often do little…’ The article does not mention how, in addition to being intrusive and painful, such attempts to treat the disease can be demeaning for patients and family members, holding out false glimmers of hope while robbing them of precious time that could be better spent outside the sterile confines of a hospital room.”
Grammarly’s assessment: “We didn’t find any plagiarism, but we found 2 writing issues.” Thankfully.
Of course, unsurprisingly, other AI programs are designed to defeat AI plagiarism detectors like Grammarly’s. For example, I came across a tool with the straightforward name, “AI Humanizer,” which promises to “Write and humanize AI articles, reports, and blogs. Bypass AI detection. All with one click.” And there are other such programs.
***
As a would-be writer, I am dismayed by how easily an original writer might be smeared as a plagiarist by a large language AI model trained on library-sized amounts of written material. After all, even its designers aren’t certain precisely how AI conducts its business. Is it possible that autonomous scrutiny could have undermined the careers of our great writers? The implications for original writers who work in the age of AI are scary.
***
In any case, this inter-Google presentation hooked me, so for laughs and giggles, I decided to put AI through its paces. First, I concocted a short scenario for a crime novel, which I fed into ChatGPT along with instructions to produce a modest, 200-page novel. After praising me for proposing a fascinating plot (I find that Chat’s basic AI model is too quick to heap uncritical praise on the user), it inquired whether I might like to see a chapter outline. When I said yes, it spat out an outline containing twenty-six separate chapters, replete with headings and summaries, all in less than one minute. This output was indeed impressive.
Next, I asked it to provide a draft of my first chapter, and it turned one out in about one minute. Despite being riddled with crime novel cliches, I must admit the feedback I got was logical to the goal of quickly “writing” a crime novel. At this speed, someone could easily pound out a couple of books per month, just as the presentation I had seen hinted. There you have it: an immediate prospect of fame and fortune, courtesy of artificial intelligence.
Of course, if you want to know how my AI-generated book ends, you’d have to buy it. Who has time for art when we’re only in it for the money?
