The AI writer

Dirt is a daily email about entertainment.

Terry Nguyen, Dirt's senior staff writer, on reframing the conversation around ChatGPT and AI from threat to tool. The second part of this essay will be published next Friday.

“For a very long time everybody refuses and then almost without a pause almost everybody accepts. In the history of the refused in the arts and literature the rapidity of the change is always startling.” —Gertrude Stein, Composition as Explanation (1926)

The billboard says “A.I. your people will love.” I spot it one night while waiting to cross the Bowery thoroughfare. I am unsure whose people exactly, as they are surely not my people, although I could be among them, another smiling face in the row of multicolored circles. The line-up is as multicultural as a Netflix cast. The ad is for Writer, an “enterprise AI writing tool” that uses natural language processing (NLP) and artificial intelligence to help users write by suggesting words and phrases to auto-complete their sentences. Writer claims to be the only writing assistant on the market that can “detect and fact-check output for you.” It promises that the “AI-generated content will sound like you, every time.”

I am suspicious of this friendly use of the second person, a stylistic quirk from the copywriting powers that be. Or maybe I’ve given Writer, which is created and owned by the Indian software company Zoho, too much credit. The copy could’ve been outsourced to some underpaid contractors found on Upwork, who hadn’t a clue of the Bowery billboard. Either way, it isn’t me they’re speaking to. Their target audience is company executives and business owners, people whose entire job is ensuring that the bottom line of a balance sheet remains green, even at the expense of people like me, the writers, corporate content creators, and creative laborers, whose jobs are increasingly at risk of being automated away.

Earlier this month, Insider announced a newsroom-wide plan to incorporate ChatGPT into journalists’ research and brainstorming process. Now, apparently, is the time for reporters to “begin experimenting with this powerful yet poorly understood new technology.” The memo, sent by editor-in-chief Nicholas Carlson, was riddled with contradictions. Employees were instructed not to insert any AI-generated sentences into their scripts or articles, but the next sentence acknowledged that this mandate may very well change. The announcement sparked confusion and outrage, although an effort was made to soften the foreboding news.

“ChatGPT is not a journalist,” Carlson affirmed. “​​You are responsible for the accuracy, fairness, originality, and quality of every word in your stories.” Rest assured, the statement seemed to say, your human labor is valuable to us. You are responsible for minding the machine—until that, too, is subject to change.

Join the private email of the preeminent fashion journalist of her generation

Line Sheet, written by Lauren Sherman, is the newest private email to launch from Puck, the ground-breaking media startup heralded by The New York Times as “The Vanity Fair for the Substack Era.”

Lauren a former top executive at Business of Fashion, writes about what everyone in fashion is talking about privately. In a recent dispatch, she dished on the current state of the fashion M&A market, with an eye on Ganni, Proenza Schouler, A.L.C. and Isabel Marant. Another edition spills the behind-the-scenes story of Law Roach, best known for styling Zendaya, Megan Thee Stallion, and Celine Dion.

Sign up today to start reading Line Sheet for free.

And check out all of Puck’s newsletters here. Plus, we've secured a personal discount for 21% off an annual plan for Dirt readers.

On April 20, Insider laid off 10 percent of its labor force, a decision resulting from the downturn in ad spending and general news consumption. The layoffs were not motivated by the integration of ChatGPT, a spokesperson said. Yet, there was no surety in the statement, no security against future job losses. Once a workplace is winnowed down, jobs aren’t likely to come back in the same capacity, even when the economy recovers.

An anxious haze settled over the internet last November when ChatGPT was released for public use. ChatGPT, as you may already know, is a large language model (LLM) trained on a massive corpus of text data that responds to written prompts from users. It can be instructed to write fiction, short blogs, poems, code, advertising copy; anything that contains words is fair game. Most of the machine’s output is, at best, mediocre and horribly monotone, but this limited performance hasn’t deterred users. The bot received about 100 million monthly active users in January, and anecdotal evidence suggests that consumer adoption has only accelerated. Further developments promise to improve ChatGPT for general use, such that its output will sound ever-more personable. The human remains, as always, the hallmark to emulate.

AI’s rapid improvements have exacerbated a collective unease among writers. The very skill that humans have developed over millennia, one that imbues us with historical consciousness, is being usurped by machines—at least, that’s the alarmist take, rehashed every week on the New York Times Opinion page. There has been a lot of talk about our “textual culture” and the future of writing, even though language models and linguistic experimentation with such tools extend as far back as the 1950s. (Aside from ChatGPT, there are many kinds of data-derived text prediction engines, distinctions that I won’t parse here.)

OpenAI’s latest update, GPT-4, was released on March 14. It can process up to 25,000 words of text and respond to both image and text queries, and has reportedly exhibited “human-level performance” on the SAT, GRE, and other academic exams. More developments are on the horizon, of course. Major tech companies, from Meta to Google, are developing products to capitalize on the AI hype. They are a series of Pandora’s boxes. For creative workers, the threat of replacement has snowballed into an identity crisis—work being the basis for most Americans’ identity, even if we’ve claimed to no longer dream of labor. But creative work is tied to a sense of purpose. It transcends the purely transactional nature of labor. I write for a living; I also live to write. The threat of AI, then, begins to feel personal, even if its implementation is strictly business.

 🌱 JOIN THE DIRTYVERSE

• Join our Discord and talk Dirt-y with us. It’s free to join! Paid subscribers have access to all channels.

• Follow @dirtyverse on Twitter for the latest news and Spotify for monthly curated playlists.

• Shop for some in-demand Dirt merch.  🍄