Is AI Just the Modern Way to Write?

Photo by Mohamed Nohassi on Unsplash

by R.K. West

According to a recent Yahoo article a majority of books in Amazon's "Success" self-help genre contain AI writing. AI books are common in other genres, too. Authors are concerned about the flood of AI-generated books that detract from the ability of real writers to sell their books.

It may be that prolific readers of genre fiction can't tell that they are reading bot-produced books, or that they don't care. Amazon permits "authors" to upload as many as three books per day, a limit that is supposedly aimed at cutting down the number of low-quality AI (and low-quality human-produced) books flooding the market.

YouTube has hundreds of videos explaining how to detect AI writing, and hundreds explaining how to make a fortune as an AI "author".

Most legitimate publishers do not want to handle AI-generated works, and have added AI-detection to the list of skills editors need. There are many automated AI detectors (which probably use AI), but they are not as reliable as we might hope. People who have studied the writing "style" of bots like ChatGPT point to certain characteristics as indicative of AI-generated texts.

Excessive use of the "em-dash". The em-dash is a punctuation mark, a long dash usually produced by typing two hyphens. it is often used much like a semicolon or parentheses, or in place of commas, to set off a clause that contains additional or explanatory material. Typically, there will be a space both before and after the dash.

John was worried about his shoes -- they often slipped off his feet while he walked -- so he secured them with duct tape.

Apparently, ChatGPT uses em-dashes to excess, and does not put spaces around them.

The words excited him--he was easily excited by words--and he began to hum. His wife--always attuned to his vocalizations--told him to calm down.

Do writers have to stop using dashes? No. If dashes are a natural part of the writer's style, they can stay, but it's best (even in a world without AI) to avoid going overboard with them.

Overuse of certain words. It seems that some "buzzwords" will surge in popularity with bots and begin appearing in texts more frequently than usual. Some of these words found in recent studies are delve, empower, elevate, and grounded. There are lists available that identify suspect words. Does this mean writers must stop using these words? No, we should feel free to use any appropriate word at any time. Even so, it is probably wise to be aware of these words and avoid over-using them.

Clichés. This should come as no suprise. Since AI is trained on existing work, it will inevitably absorb cliché phrases. If someone is "inexorably drawn into a web of deceit" and finds herself "shouting into the void," it could be AI, or it could just be an unimaginative human writer.

Formulaic structures. AI systems like to follow patterns. They will repeatedly show contrast by saying, "It isn't just X. It's also Y." or "Not A. Not B. Just C." Apparently, they also like the "rule of three," whereby lists must always contain three elements. This is unfortunate, because to many human writers, a list of three just "sounds" better than two or four.

Dishonesty. Bots have been caught writing about newsworthy events that never actually happened. They have cited fake sources, including imaginary studies and articles or books that don't exist. They have used incorrectly attributed quotations and have referred to writers and experts who do not exist, or whose expertise is in a different field. They use made-up statistics without attribution.

Studies show that 87% of novels on the New York Times best-seller list are 100% AI-written. As microbiologist Joan Didion once said, "I have seen the future of literature, and it is bot."

Sadly, humans sometimes do these things, too.

Bad Writing. AI writing frequently just seems "bad," however one might define that. It is often unimaginative, lacking rhythm, dependent on exaggeration and cliches, and repetetive. There may be nonsense sentences, full of words, apparently grammatically correct, but devoid of meaningful information. Of course, bad writing is often produced by humans, whose creativity makes it possible to exceed AI in unreadability.

Even without referring to a checklist, experienced readers can identify AI writing with a high degree of certainty.

It is hard to imagine that anyone using AI to generate texts could call themselves an "author" without a twinge of guilt. Do the purveyors of these word collections (based on plagiarism) actually feel creative? Is it possible to take pride in having accomplished nothing beyond distributing mediocre word slop?

Aside from poor quality, there are many practical reasons to not use AI, especially for those who hope to have a professional writing career.