As soon as I sit down to write, I feel compelled to scrub my bathtub and reorganize my filing cabinet — the most tedious chores suddenly become more appealing than the task at hand. Writing can feel so daunting that we’ve invented the term writer’s block to describe the unique sensation of its challenge, and we debate whether the ability to write well is learned or simply innate. The work requires long stretches of intense focus and undivided attention, and doing it well usually involves a prolonged process of revision. For many of us, writing feels like one of the most burdensome activities we can do.
Generative AI tools like ChatGPT offer the seductive possibility that we can optimize this laborious process. But while it can clearly optimize the time and effort of writing, ChatGPT cannot necessarily optimize writing quality. The program produces highly competent prose that usually passes as human-generated, but so far, the quality of its writing — beyond the novelty of being authored by an algorithm — is mostly unremarkable.
At the University of California, Los Angeles, where I teach writing, the common sentiment among faculty is: “Sure, ChatGPT can write — but it can’t write well.” Some professors caution students against using the tool by appealing to their egos: “You could use AI to cheat on your essay, but do you really want a C+?”
Others, recognizing that AI tools will characterize the working world into which students will graduate, are beginning to allow their use in constrained ways, framing them as automated writing tutors or advanced grammar-checking tools. But even AI enthusiasts tend to advise students to maintain authorial control by editing any AI-generated output for accuracy, style and sophistication.
The flat, conventional feel that characterizes most AI-generated writing stems from the predictive nature of the algorithm. Trained on vast databases of human texts, from books to articles to internet content, programs such as ChatGPT, Bard, Bing, and Claude function like sophisticated autocomplete tools, identifying and predicting phrase patterns, which makes their output feel somewhat predictable, too.
But does predictable writing necessarily mean bad writing? When we talk about good writing, what exactly do we mean? As we explore new applications for large language models and consider how well they can optimize our communication, AI challenges us to reflect on the qualities we truly value in our prose. How do we measure the caliber of writing, and how well does AI perform?
In school, we learn that good writing is clear, concise and grammatically correct — but surely, it has other qualities, too. Perhaps the best writing also innovates in form and content; or perhaps it evokes an emotional response in its readers; or maybe it employs virtuosic syntax and sophisticated diction. Perhaps good writing just has an ineffable spark, an aliveness, a know-it-when-you-see-it quality. Or maybe good writing projects a strong sense of voice.
But then, what makes a strong voice, and why does ChatGPT’s voice so often fall flat?
The Value Of Human Error
“The Elements of Style,” the classic reference book on writing by William Strunk Jr. and E.B. White, lays out a series of concrete rules. To write well, the authors say, you should abide by certain conventions, such as grouping your sentences into single-topic paragraphs. You should adhere to certain grammatical rules, like, “Do not join independent clauses by a comma.” You should “omit needless words” and write in an efficient, organized, streamlined manner.
These rules take effort for any human writer — we all miss the occasional comma splice, use a few more words than necessary or bury our main point in the middle of a paragraph. ChatGPT, by comparison, rarely makes rhetorical moves that stray from Strunk and White’s conventions unless instructed to do so, and the speed with which it spews forth efficient, grammatically correct sentences is impressive, unsettling and perhaps mildly humiliating to us error-prone human writers. For teachers trying to catch cheating students, the total absence of typos and grammatical flubs is often what raises suspicions.
We seem to tolerate and even expect a certain amount of idiosyncrasy in our writing, and the conventions themselves can be murky and variable— the Oxford comma, for instance, maintains a devoted cult of enthusiasts even while some style guides discourage its use, and languages like African American Vernacular English have their own coherent grammatical structures that differ from those of so-called standard American English. Conventions can also evolve over time — we now commonly treat “they” as a singular pronoun when a short time ago it was exclusively plural.
Seguir leyendo: Noéma