As news outlets begin experimenting with Artificial Intelligence to help write their stories, the tech news site Gizmodo this week was reminded of the importance of editors in overseeing the process.
On Wednesday, about ten minutes after employees at the outlet were told AI-assisted bots would be generating news stories for the site, an article by "Gizmodo Bot" was posted, listing all the "Star Wars" films and TV shows in chronological order — and it was riddled with errors, The Washington Post reported.
Gizmodo Deputy Editor James Whitbrook told the Post he counted 18 "concerns, corrections, and comments" he had about the story — including that the AI bot inaccurately listed the chronology of the TV series "Star Wars: The Clone Wars," omitted "Star Wars: Andor" and the 2008 Clone Wars film from the listing, and included formatting errors and repetitive descriptions of the George Lucas universe.
"I have never had to deal with this basic level of incompetence with any of the colleagues that I have ever worked with," Whitbrook told the Post, adding that if chatbots "can't even do something as basic as put a 'Star Wars' movie in order one after the other," then readers cannot trust it to report accurate information.
The story was one of five articles published using Google's Bard and OpenAI's ChatGPT technologies as the outlet pilots new AI initiatives, a representative of Gizmodo told Insider. The representative would not comment on any editorial review process that the articles were subjected to.
In an email sent to Gizmodo employees on June 29 and forwarded to Insider by a Gizmodo representative, Editorial Director Merrill Brown said the articles are "an early, modest test. There will be errors, and they'll be corrected as swiftly as possible. There will be future pilot projects, as well. These will all be designed to complement our journalism and give our editorial teams new tools to serve our audiences."
Gizmodo employees were quick to register their displeasure with the articles, reacting with thumbs down and poop emojis at a message from Brown in the company's Slack channel announcing plans to "thoughtfully gather and act on feedback" about the AI pilot program, the Post reported.
The publication's union sent an open letter to company leadership urging them to reconsider their stance on AI-generated articles, saying, "computer-generated garbage undermines our ability to do our jobs, erodes trust in us as journalists, damages our brands, and threatens our jobs."
The GMG Union, representing authors from The Onion, Jezebel, and The Root, publicly urged readers not to click on AI-written articles, arguing that the stories are "unethical and unacceptable."
Representatives for the Gizmodo Union did not respond to Insider's requests for comment.
Gizmodo reporters weren't the only ones angered at the flub — readers, too, expressed their dissatisfaction with the AI creation.
"Posting AI spam is contrary to the reasons people read your various properties. If you do this I and certainly many other longtime readers will bail out, leaving your ad revenue in the toilet," one reader wrote on Twitter.
Another called the articles "perfect examples how some executives clearly do not understand their customers (readers), have only a very loose association with integrity, and solely focus on cost and profit instead of quality and value.
Claire Lower, a reporter with LifeHacker, posted a screenshot of an email draft to the executives spearheading the decision — Brown, CEO Jim Spanfeller, and Deputy Editorial Director Lea Goldman — which read: "Your legacy is one of destruction. Of taking something well-loved and stripping it of everything that made it interesting and unique, and it has been heartbreaking to watch."
"Your only motivation is generating profit, and you're not even good at it," Lower added.
AI-generated content has become a hot-button topic in newsrooms nationwide.
Job security for reporters remains on shaky ground. There have been massive layoffs across the media landscape this year, including at BuzzFeed, NPR, and Vice.
At the same time, outlets — including Insider — have announced new initiatives to experiment with AI, with editorial procedures in place meant to prevent errors from being published.
