With artificial intelligence stepping in, who needs writers?
M❚❚A report by CBS News asserts that ChatGPT could ably perform the work of writing government legislation, legal documents, or computer code, as well as the less specialized tasks of composing routine emails, press releases and advertising copy. The future of writing for pay seems shaky.
any years ago, in the last century, I was a high school English teacher. It was not a good fit for me, but I am thinking today of a young man, a junior, who had done exactly no work in my class all semester. I was happy when he finally turned in an assignment, a short story that was very good. In fact, it was awfully good, and my heart sank when I typed his first line into Google and the whole thing appeared, published by an actual writer. I gave him an F, sent him to the office, called his parents, the whole deal.
I felt I’d failed, too: My teaching had not kindled his creativity. That student is an adult now, possibly raising little plagiarizers of his own. His kids are going to love ChatGPT.
Writers are writing about ChatGPT, buzzing like bees about the new artificial intelligence (AI) program that, according to its designers, “interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.”
Yikes. Forget the worry that ChatGPT will put us writers out of jobs. It sounds like ChatGPT would make an excellent parent.
But the part about sending writers to the unemployment line is troubling. A report by CBS News asserts that ChatGPT could ably perform the work of writing government legislation, legal documents, or computer code, as well as the less specialized tasks of composing routine emails, press releases and advertising copy. The future of writing for pay seems shaky.
Another avenue of AI encroachment, as one can imagine and as students have already imagined, is the manufacture of high school essay assignments, which will be a breeze for ChatGPT. Educators are concerned, because ChatGPT doesn’t just copy something found online, as my long-ago student did. It produces an original outlay by scouring the internet for information and examples and then synthesizing the results of its search. Almost instantly.
The ChatGPT program has only been available to test since November. I recently checked it out for myself. The first time I tried to join, I was informed onscreen that the website was too busy to accommodate my registration. I was then offered 10 jokes of dad-joke caliber that ChatGPT had written to excuse its unavailability. The next day, I got to log in. I wanted to broach a topic that would be similar to an English class essay. So I asked ChatGPT, “What do the daughters of ‘King Lear’ symbolize?”
The answer was fast and perceptive. ChatGPT listed the three daughters by name, zeroed in on their dominant qualities, and then noted, “These characterizations are used to explore themes such as filial duty, the consequences of selfishness and cruelty, and the search for truth and redemption.”
Holy cow. I was impressed. I was even charmed by the Oxford comma.
Then I asked, continuing our dialogue, “Why are they daughters and not sons?”
ChatGPT expounded on the particulars of inheritance law in Elizabethan England and the gender dynamic between father and daughters. In response to my final question, I received a list of adaptations and retellings of the “King Lear” story over time, both literary and cinematic.
And that gave me everything I’d need to craft a typical 500-word high school essay. ChatGPT didn’t exactly write it for me, but it could have upon request. It did hand me easy and elegant research. With a bit of copying and pasting, I’d surely get an A.
But to circle back to the predicted demise of professional writers: As ChatGPT and other AI programs become more widespread, will writers become obsolete?
This writer hopes that worry is being overblown. By writers. ChatGPT can imitate the style of a writer, but it cannot replace the writer. Human tasks can be replicated and made easier, but humanity is unique to each of its beings. An AI tool that uses only prior knowledge cannot ignite the spark of new creation. A bot may be clever, but it lacks a soul. Which is where the best writing is lit.
And I am heartened by the work of Edward Tian, a senior at Princeton University. Mr. Tian has developed GPTZero, an online tool to detect ChatGPT cheating. Purporting an accuracy rate of 98 percent, it is free to teachers. “Your text is likely human generated!” its analysis assures the user. Or not.
Mr. Tian may seem like a total narc to other students, but teachers will be grateful for this aid in counteracting a new plagiarism headache. The industrious Mr. Tian demonstrates that no matter how smart AI becomes, humans will rise to the challenge to outsmart it. Or at least endeavor to live in uneasy peace with it. It seems we love our technology too much to part with it.