A laptop that says "AI assistance mode: off."
In an opinion column titled “Let the Bond Market Speak” that ran recently in The Wall Street Journal, billionaire and former hedge fund manager Stanley Druckenmiller argued that Treasury Secretary Scott Bessent should not intervene in the bond market. After the op-ed was published, Claudia Sahm, the chief economist for New Century Advisors, used Pangram, an AI detection tool, to review the piece. It found that “100% of this text is AI,” according to Notus, which covers government and politics.
When Notus interviewed Druckenmiller, he didn’t deny it. Quite the opposite. “Of course I used AI,” he said. “There’s a reason I moved from an English major to being an economics major,” Druckenmiller explained. “I’m not embarrassed by it…I write everything using AI now for the same reason I use a calculator when I do math problems.”
Perhaps more surprisingly, Paul Gigot, the editorial page editor for The Wall Street Journal, vigorously defended Druckenmiller, writing in a statement that “AI is a fact of modern life.”
“People will use it to assist in their work and their writing, including with research, checking grammar, editing and more,” he added. “The question for us is whether what we publish from contributors reflects an author’s original argument, and if the author has the standing and credibility to make it. In Stan Druckenmiller’s case, we have had a relationship with him for many years, and nobody can doubt that his op-ed is his genuine opinion.”
You might remember a more local hullabaloo, when a story riddled with AI hallucinations made it to the front page of the Sunday edition of the Wisconsin State Journal. Editors there eventually pulled the error-ridden piece, replaced it with a new report and fired the reporter who authored the original version. The State Journal never shared what exactly went wrong or how processes might be adjusted to avoid a similar problem in the future.
Seeking context, I reached out to Tomás Dodds, who had just joined the faculty at UW-Madison, launching the Public Tech Media Lab in the process. His take on the WSJ mishap and similar ones (remember this doozy? “A.I.-Generated Reading List in Chicago Sun-Times Recommends Nonexistent Books”) is that they reflect more of a systems lapse than an employee misstep.
“What we are seeing here is organizations’ failures in training journalists,” Dodds told me, adding that among the services his lab planned to offer was AI policy training. “Sign me up,” I recall thinking at the time.
About a month later, Dodds came to our offices at the coworking space StartingBlock, where Isthmus is a member, for the first of two workshops. In the first session, staff discussed ways they currently use AI, including for headline and art brainstorming and recording and transcription of interviews. In a second session, we talked more specifically about what we’d like our policy to look like. Dodds later drafted a policy, which we edited, incorporating some features from local news organizations across the country.
A few highlights: While staff or contributors may use AI for research, every fact must be independently verified by the person using the AI — the writer is responsible for accuracy. Also, “AI is not used to write or edit content. We do not use AI for fact-checking or in place of our own editorial judgment.”
The purpose of the guidelines is twofold: They guide the work of our staffers and contributors and signal our principles to our readers. We feel they are important for quality control and transparency. Our AI policy is now posted on our website. Given the speed of advancing technology, it is something that will likely need frequent updating.
Have questions about our AI policy? AI use in general in the milieu of local news? Join us for “AI & Journalism: A community conversation with Madison journalists” on Sept. 17 at Cargo Coffee, 750 E. Washington Ave., from 6-8 p.m.
Angela Russell, former chief diversity officer at TruStage, creator and host of the Black Oxygen podcast, and the recent founder of The Undercurrent, a consulting practice focused on employee engagement and social impact, will moderate a panel discussion with yours truly, Michael Wagner, the director of UW-Madison’s journalism school and the Center for Communication and Civic Renewal, and Omar Waheed, reporter with Madison365.
I asked Waheed about how AI might be affecting the work his news outlet does. He says it really hasn’t.
Waheed, who also serves as managing director of the Los Angeles Center for Investigative Journalism, attended a series of trainings last year on AI, in part to help the board of directors at Madison365 identify whether they would pursue any investment in AI. “Ultimately we decided ‘no,’” he says.
“We’re not going to write anything using AI,” Waheed says. “We’re not going to use AI photos.” Waheed does use AI transcription services and says he can see the value in data visualization tools powered by AI and benchmarking tools to identify some “hot button issues” during recordings of long city council meetings.
News organizations are being enticed to consider AI in their workflow. An executive with the AI operating system Nota made a pitch for its use in an email with the subject line, “Spend more of your time on stories that matter.”
“Nota Studio can take any media input; a PDF, a release, an audio or video file, a url, and produces a drafted article in your voice.” It then forwards the result to your content management system “already optimized.”
The executive, Evan Young, touted the program’s success: “One regional partner cut pre-publish production time by 70% and saw page views rise 39%. They saved a lot more than a few hours. They generated net new readers and revenue.”
The company does acknowledge the need for human intervention, at least at some point: “While our AI is pretty smart, it’s not a magic wand. It can’t write the content for your breaking story in one fell swoop or brew the perfect cup of coffee (although we’re working on it). Please use our tools responsibly and consult with your customer success lead for assistance.”
Hope you can join us on Sept. 17. Please use the QR code to RSVP. There will be light refreshments, generously donated by co-host Cargo Coffee, and time to socialize with attendees. Please come with questions. I think we’re in store for a spirited conversation.
