6 July 4-10, 2024 miaminewtimes.com | browardpalmbeach.com New Times | music | cafe | culture | Night+Day | news | letters | coNteNts | MONTH XX–MONTH XX, 2008 miaminewtimes.com MIAMI NEW TIMES | MUSIC | CAFE | FILM | ART | STAGE | NIGHT+DAY | METRO | RIPTIDE | LETTERS | CONTENTS | ROBO-REPORTING News site uses AI writers with fake names for Miami coverage. BY NAOMI FEINSTEIN D o you know Hoodline reporter Carlos Mendez? How about Ethan Anderson, Alec Novak, or Livi Miller? You probably don’t because they are not real people. Although the names are featured in bylines all over Hoodline’s Miami news website cov- ering local politics, weather, food, and crime throughout South Florida, their articles are products of artificial intelligence (AI). As our friends at the Dallas Observer re- ported, Hoodline — which covers 25 cities across the U.S. including Dallas, San Fran- cisco, Chicago, Los Angeles, Tampa, Wash- ington D.C., and Atlanta — uses AI-generated articles with bogus names to simulate jour- nalism. The company reportedly started out using fabricated bios and headshots, though those features have since been removed, leav- ing only the fake names. The company’s use of AI has played into fears that artificial intelligence-written arti- cles will jeopardize the future of journalism and fuel the rapid spread of misinformation if the practice remains unchecked. Hoodline announced in April that it would place an “AI” label next to the names of the “writers” to denote articles created with AI- generated content. Odd Takes One example of irregularities in the AI-gener- ated news appears in Hoodline’s January arti- cle about the Miami-Dade State Attorney’s Office’s decision not to press charges against a cop who fatally shot local resident Jaime Ro- bles during a Miami Springs hotel raid in 2022. The Hoodline article cites New Times, which broke the story on the morning of Jan- uary 5 after obtaining a state attorney’s memo about the case and completing hours of re- search about the incident. As New Times’ cov- erage noted, a tactical unit carried out the raid after detective Cesar Echaverry was killed in a shootout while pursuing a suspect from a Dania Beach armed robbery. The Hoodline AI-generated article ap- peared to make unusual leaps in logic, includ- ing a line that says Robles was “mistakenly identified as connected” to the armed robbery. While Robles was not present during the shootout with Echaverry, the memo notes that the suspect was staying in Robles’ hotel room, kept a gun there, used Robles’ car, and that police were investigating whether Robles had a role in the robbery. Robles’ family denied his participation in the theft; still, it’s unclear how the Hoodline article concluded that, as a matter of fact, he was not “connected.” The Hoodline piece also had some odd phrasing that signaled that it was created by an AI tool rather than under the hand of a flesh-and-blood writer. “This tragedy unfolds as a narrative bereft of clear heroes and villains, the lines between law, order, and the safeguarding of innocent lives blurred as we continue to piece together the events of that fateful day,” the article waxes in robotically poetic fashion. High Output The Hoodline Miami site shows that their “reporters” are churning out content at a diz- zying pace. The website is filled with press release coverage, weather updates, crime re- ports, and coverage of missing people. Original reporting, interviews, scoops, and content requiring investigative work are gen- erally absent from the recent Miami coverage. Here is just a quick sample of stories writ- ten by their team of AI reporters: • “Fort Lauderdale Offers Sandbags for Flood Prevention as Hurricane Season Approaches.” • “Fort Lauderdale’s Holiday Park Ex- pects Bustling Event Crossover with 2,000 Attendees This Friday.” • “Texas Man Found Deceased Follow- ing Standoff with Port St. Lucie Police at Shopping Plaza.” • “Broward Sheriff’s Office Ramps Up Search for Missing Fort Lauderdale Air- port Passenger Jean Claude Beauvil” The four AI bylines for the Miami outlet published more than 100 stories from Monday, June 17 through Saturday, June 22 alone — few journalists can be that productive unless they are wired on Adderall and not sleeping. Hoodline’s Case for Fake Bylines Hoodline, whose motto is “in-depth report- ing about your home area,” used to rely on human journalists when it was founded in San Francisco in 2014, according to CNN. Impress3 says that when it bought Hood- line in 2020, the operation was “on the brink of being shut down.” The company shifted to a model where it employed “junior-level jour- nalists to use various tools,” including AI, to “reblog police press releases and create reli- able, simple news stories.” The outlet transi- tioned to primarily artificial intelligence-generated news as it expanded its coverage nationwide. The bottom of its website includes a small “Use of AI Disclaimer.” “Our stories are cultivated with a human- centric approach, involving research and edi- torial oversight,” the disclaimer reads. “While AI may assist in the background, the essence of our journalism — from conception to pub- lication — is driven by real human insight and discretion.” Impress3 CEO Zachary Chen said that ar- tificial intelligence allows Hoodline to pro- vide reporting in news deserts. He maintained that “dozens of editors as well as dozens of journalist researchers” are on full- time staff. He also told Nieman Lab that the AI personas were created by AI at random along with their beats and cities. Hoodline is certainly not the only outlet to use AI for its content — Sports Illustrated made waves last November for its AI reporters and fake bios and headshots. SI deleted the material after news broke about its use of AI. [email protected] Hoodline uses bogus names on AI- generated news articles. Image by Moor Studio/Getty Images | METRO | ▼ MIAMI ABOUT THAT AD... NEW TIMES EDITOR TOM FINKEL: “TO PUT IT BLUNTLY, THE AD IS ANTITHETICAL TO WHO WE ARE.” BY MIAMI NEW TIMES STAFF A s Axios reported late last month, Miami New Times published an ad containing a racial slur on the back cover of our June 13 print issue. New Times publisher Adam Simon explained to Axios that the ad came in after the press deadline and was not reviewed before publica- tion. “Naturally, had I seen it, which I should have, I would not have let it run as is,” Simon said. “The bottom line is that it was gross and unacceptable oversight by our local leadership and I take 100 percent responsibility for allowing that language to run in an ad.” To prevent anything like this from happening in the future, the advertising department is im- plementing a failsafe system to ensure that every ad is reviewed before publication. New Times editor-in-chief Tom Finkel says that although his newsroom had nothing to do with the ad that ran, the editorial staff was ap- palled at its misleading and harmful language because it reflects on the paper as a whole. “To put it bluntly, the ad is antithetical to who we are,” Finkel says. “Mutual trust is at the core of our ethos. We believe that you trust us to dig into topics without an agenda and to share with you what we’ve learned — whether it’s good news or bad. We, meanwhile, trust that you understand that our aim is not to tell you what or how to think, but rather to make you feel more connected to the place we all live — for better or for worse.” To that end, Finkel notes, “We don’t publish a story — any story — unless we’re prepared to stand behind it. Of course, everyone makes mis- takes, and reporters and editors are no exception. When we get something wrong, we correct it.” The offensive ad was submitted by Blacks for Trump, a group whose exploits New Times has covered in the past. “We want to assure you that New Times will no longer be accepting advertisements from this client,” says Voice Media Group editorial director Chelsey Dequaine-Jera- bek, adding that Miami’s production team is go- ing through its online flipbook archive to remove any other offensive advertising content from PDF versions of print editions. “Through this experience, senior leadership across business and advertising departments have agreed on a shared commitment to raising standards and learning from these mistakes,” Dequaine-Jerabek says. “We have rewritten protocols and strengthened proofing workflows and will continue to look deeply inside our or- ganization to ensure these types of failures cease to exist.” On the heels of Juneteenth, Dequaine-Jera- bek says she’s deeply disappointed to be writing to readers about this topic. She says New Times always has a choice in running the ads it runs. It can choose not to work with clients who wish to run ads that contain racial slurs. “We want to make it clear to our readers that this instance is the furthest from who New Times is and who we will be moving forward,” she says. “We have a responsibility to our readers, our community, our clients, and our employees. This ad violated that responsibility.” [email protected] “WHEN WE GET SOMETHING WRONG, WE CORRECT IT.”
