WSJ and AI: Another Line Crossed in AI-Generated Writing

Last week, The Wall Street Journal published a guest op-ed piece titled "Let the Bond Market Speak." The author was Stanley Druckenmiller, a prominent billionaire investor. But soon after publication, the focus shifted from the state of the American economy to what was not included on the byline: the AI program that wrote the piece.

The column had enough AI flags that it, pardon the pun, prompted readers and reporters to investigate further. (The first paragraph closed with a common AI trifecta: colon-introduced clause, it's not X it's Y construction, and an em dash, all in one sentence.) When asked about his use of AI to craft an editorial for a major American newspaper, Druckenmiller was not shy. He said he used an AI program the same way someone would use a calculator to solve a math problem.

Two days later, WSJ's response dragged the AI writing debate onto new terrain. In a column of his own titled "Much Ado About Artificial Intelligence," Paul Gigot, editor of the Journal's editorial page, defended the expert's use of AI and blamed "the press" for stepping on Druckenmiller's financial insight. Contributors are free to submit AI-generated columns and essays, he said, even if the Journal's opinion writers themselves can only publish original work.

After reporting on the story, The Atlantic published a parody column written by AI and styled in Gigot's dismissive, condescending tone.

What does all that mean? It means another ragged line has been crossed in the evolution and acceptance of AI writing. A major newspaper approves publication of AI-generated writing under the condition the contributor is accomplished enough in their field to prompt it. Or edit it? Or fact-check it? Who knows.

Unfortunately, without any semblance of leadership on AI, each entity is forced to come up with its own rules, whether it's a newspaper or magazine, a professional editor or self-published writer. Editors are drawing their own lines in the sand: Do I take on writing projects or manuscripts that have been assisted by AI? Do I refuse to touch them? Do I force clients to disclose whether or not they've used any type of artificial assistance? On what grounds?

The next logical question is at what point do newspapers and other publications accept AI works that mimic an expert's insight without actually coming from the expert at all. (Write a 2,500-word essay on the looming Major League Baseball labor dispute from the perspective of Bob Costas and Peter Gammons.)

It likely won't be long before we find out.