• | 9:00 am

As AI encroaches on journalism, I’m concerned—but not panicky

Media outlets keep adopting AI without considering the consequences. But I don’t see the human element disappearing from the craft—and it might even become more crucial.

As AI encroaches on journalism, I’m concerned—but not panicky
[Source photo: Clay Banks/Unsplash]

Thanks for reading Plugged InFast Company’s weekly dispatch from the world of tech. If a friend or colleague forwarded this edition to you—or you’re reading it on FastCompany.com—you can check out previous issues and sign up to get it yourself every Wednesday morning. I’m always eager to hear from you with feedback and ideas: Shoot me an email at hmccracken@fastcompany.com.

On November 27, news broke that Sports Illustrated had stomped on one of the biggest rakes in journalism. As reported by Futurism’s Maggie Harrison, the venerable sports media brand—now managed by a company called Arena Group—had been publishing articles written by imaginary people with AI-generated head shots. Furthermore, a source at SI told Futurism that at least some of the stories had been generated by AI.

After Futurism noticed the pieces, SI deleted them from its site, blaming the mess on a partner called AdVon Commerce and terminating the relationship. SI acknowledged that the credited authors were fictional but said that AdVon had denied that the articles were AI-generated. Not that Arena Group was steadfastly opposed to letting bots write for its sites: Earlier in the year, it issued a press release touting the use of AI by another one of its brands, Men’s Journal. But that had blown up in the company’s face, too: Futurism pointed out serious factual errors in the first AI-written Men’s Journal story.

Arena Group is just one of the alarming number of media companies that got irrationally exuberant over AI this year. Others have included Cnet, Gannett’s Reviewed, and G/O Media, the publisher of Gizmodo and AV Club. The details of each kerfuffle varied. But in each case, it originated with a Wile E. Coyote super genius plan to scale up content without paying human beings to create it—and ended in humiliation.

Now, as a human who’s spent my career getting paid for writing stuff, I’m hardly a disinterested party here. I hope to continue stringing words together into articles, a goal that might be complicated by the media industry’s interest in computerizing the process.

But even on a less self-serving level, it pains me to think that the future of the business that employs me could lay in synthetic verbiage rather than lovingly handcrafted work. Thinking of journalism as a calling—not a cost center subject to ruthless automation—is one of the things that motivates me to wake up each morning. (For the record, Fast Company has no plans to publish anything composed by a bot, which comes as a relief in every possible way.)

Of course, it’s easiest to turn your nose up at AI-generated text when it’s drearily written, bereft of insight, and riddled with errors, as is presently a common scenario. But what happens when the tech improves enough that embracing it might not be a grave mistake? I certainly don’t want to be the 21st-century equivalent of the New York Times tech writer who confidently declared laptops and Windows to be passé in the mid-1980s, well before they took off. (And who, I should add, was not a dunderhead—just a smart guy who made the mistake of predicting the future based on his own predilections and where the tech stood at the time.)

So let’s assume for the moment that computers will get better at putting together articles that are at least as publication-worthy as something a competent human journeyman might have written. And let’s further declare that they’ll be clearly labeled as AI-generated rather than snuck onto sites in a misleading fashion. That would still present fundamental conundrums for journalism. For instance, what would it mean for early-career reporters if the kind of stories they would most likely be assigned would go instead to a bot?

Overall, though, I can live with the prospect of machines playing a growing role in media. First, the conventional wisdom that turning the least rewarding work over to computers lets humans focus on more rewarding pursuits might turn out to be true. No publication will turn its best ideas over to an AI: They’ll start with the ones that feel largely obligatory. That really might free up time, money, and brain cells for higher-stakes endeavors.

Secondly—and I’m trying to be optimistic here!—writing by experienced humans might stand out in a surging sea of commoditized, algorithmic content. The yen to improve the world is a powerful component of the best journalism. So is a reputation that encourages interview subjects and other sources to trust you. And it still feels like AI might be years away from being a delightful, thought-provoking prose stylist rather than a dutiful-but-unremarkable wordsmith. If flesh-and-blood journalists feel under pressure to make their work as special as possible, we’ll all benefit.

Lastly, as long as human journalists remain in charge and behave responsibly, technology can facilitate better work. Already, I run most of what I write—including this very column—through Grammarly. I accept some of its AI-powered suggestions, reject many others, and expect to call on additional AI assistants as they prove their worth. That’s a far different scenario than the recent controversies involving AI-generated content, most of which seem to stem from suits on the business side of media companies imposing content strategies on the editors. That’s always a risky proposition, whether or not computers are involved.

It pays to remember: This year’s dubious achievements in AI-produced content didn’t occur because the technology isn’t ready for prime time. They happened because humans in the media business failed to think through the implications of their actions. While I expect additional outlets to fall victim to AI’s siren call, I do think that a decent percentage will proceed with noble intentions and common sense. That doesn’t guarantee any happy outcomes, but it’s the best way to gird ourselves for the never-ending tumult ahead.

Bonus material: As an experiment, I asked ChatGPT and Bing’s Copilot to compose essays on this newsletter’s topic based on this prompt:

Write a lively 900-word essay about whether human journalists should be worried about increasing use of AI to generate articles, some of which has resulted in media outlets publishing embarrassingly shoddy work. Begin the story by providing some specific recent real-world examples of these outlets’ mishaps and their consequences.

ChatGPT seems to have made up its example of an AI mishap; Bing managed to provide some real ones. Both bots touched on some of the same points I did, but didn’t leave me too stressed over the prospect of being rendered irrelevant. See what you think.

  Be in the Know. Subscribe to our Newsletters.


Harry McCracken is the technology editor for Fast Company, based in San Francisco. In past lives, he was editor at large for Time magazine, founder and editor of Technologizer, and editor of PC World. More More

More Top Stories: