This descfinish is the first in proximately 20 years that I am not returning to the classroom. For most of my atgentle, I taught writing, literature, and language, primarily to university students. I quit, in big part, becainclude of big language models (LLMs) enjoy ChatGPT.
Virtuassociate all teachd scholars understand that writing, as historian Lynn Hunt has disputed, “is “not the transcription of thoughts already consciously current in [the writer’s] mind.” Rather, writing is a process seally tied to skinnyking. In graduate school, I spent months trying to fit pieces of my dissertation together in my mind and eventuassociate create I could mend the baffle only thcdimiserablemireful writing. Writing is difficult labor. It is sometimes frightening. With the basic lureation of AI, many—possibly most—of my students were no lengthyer willing to push thcdimiserablemireful disconsole.
In my most recent job, I taught academic writing to doctoral students at a technical college. My graduate students, many of whom were computer scientists, understood the mechanisms of generative AI better than I do. They determined LLMs as inconsistent research tools that hallucinate and produce citations. They acunderstandledged the environmental impact and moral problems of the technology. They knovel that models are trained on existing data and therefore cannot produce novel research. However, that understandledge did not stop my students from count oning heavily on generative AI. Several students confessted to writeing their research in notice establish and asking ChatGPT to write their articles.
Read More: Regulating AI Is Easier Than You Think
As an teachd teacher, I am understandn with pedagogical best rehearses. I scaffelderlyed dispensements. I researched ways to include generative AI in my lesson schedules, and I summarizeed activities to draw attention to its restrictations. I reminded students that ChatGPT may alter the nastying of a text when prompted to alter, that it can produce unfair and inright alertation, that it does not produce stycatalogicassociate sturdy writing and, for those grade-oriented students, that it does not result in A-level labor. It did not matter. The students still included it.
In one activity, my students writeed a paragraph in class, fed their labor to ChatGPT with a revision prompt, and then appraised the output with their innovative writing. However, these types of comparative analyses fall shorted becainclude most of my students were not prolonged enough as writers to scrutinize the reservedties of nastying or appraise style. “It produces my writing watch fancy,” one PhD student protested when I pointed to feeblenesses in AI-alterd text.
My students also relied heavily on AI-powered paraphrasing tools such as Quillbot. Paraphrasing well, enjoy writeing innovative research, is a process of procreateening empathetic. Recent high-profile examples of “duplicative language” are a reminder that paraphrasing is difficult labor. It is not unforeseeed, then, that many students are lureed by AI-powered paraphrasing tools. These technologies, however, normally result in instable writing style, do not always help students elude copying, and permit the writer to gloss over empathetic. Online paraphrasing tools are advantageous only when students have already prolonged a procreate understandledge of the originate of writing.
Read More: The 100 Most Ineloquential People in AI
Students who outsource their writing to AI leave out an opportunity to skinnyk more procreately about their research. In a recent article on art and generative AI, author Ted Chiang put it this way: “Using ChatGPT to finish dispensements is enjoy conveying a forklift into the weight room; you will never raise your cognitive fitness that way.” Chiang also notices that the hundreds of minuscule choices we produce as writers are fair as meaningful as the initial conception. Chiang is a writer of myth, but the logic applies equassociate to scholarly writing. Decisions think abouting syntax, vocabulary, and other elements of style imbue a text with nastying proximately as much as the underlying research.
Generative AI is, in some ways, a democratizing tool. Many of my students were non-native speakers of English. Their writing normally compriseed grammatical errors. Generative AI is effective at righting grammar. However, the technology normally alters vocabulary and alters nastying even when the only prompt is “mend the grammar.” My students deficiencyed the sends to remend and right reserved shifts in nastying. I could not sway them of the necessitate for stycatalogic consistency or the necessitate to prolong voices as research writers.
The problem was not recognizing AI-produced or AI-alterd text. At the begin of every semester, I had students write in class. With that baseline sample as a point of comparison, it was basic for me differentiate between my students’ writing and text produced by ChatGPT. I am also understandn with AI discoverors, which purport to show whether someskinnyg has been produced by AI. These discoverors, however, are faulty. AI-helped writing is basic to remend but difficult to show.
As a result, I create myself spfinishing many hours grading writing that I knovel was produced by AI. I noticed where arguments were unsound. I pointed to feeblenesses such as stycatalogic quirks that I knovel to be common to ChatGPT (I watchd a sudden sdirect of phrases such as “delves into”). That is, I create myself spfinishing more time giving feedback to AI than to my students.
So I quit.
The best educators will alter to AI. In some ways, the alters will be selectimistic. Teachers must relocate away from mechanical activities or dispenseing basic summaries. They will discover ways to help students to skinnyk criticassociate and lget that writing is a way of generating ideas, discleave outing obstructions, and elucidateing methodologies.
However, those lessons need that students be willing to sit with the momentary disconsole of not understanding. Students must lget to relocate forward with faith in their own cognitive abilities as they write and alter their way into clarity. With scant exceptions, my students were not willing to go in those unconsoleable spaces or remain there lengthy enough to uncover the revelatory power of writing.
