Having spent a quarter-century as a photojournalist with the St. Louis Post-Dispatch, last year David Carson decided it was time for a break. However, instead of vacation, the Pulitzer Prize-winning photographer used the time off to study the collision between journalism and artificial intelligence.
Carson, a 2025 John S. Knight Journalism Fellow at Stanford University, said he became concerned about AI after encountering a string of images that showed how falsehoods can spread quickly — from purported news photos of Donald Trump’s arrest to fake images of smoke rising from the Pentagon.
Of course, creating images of fake events and people isn’t new.
“People could do this for years with Photoshop,” Carson said, “but creating high-quality fakes really took a lot of technical skill. What Dall-E and these AI image generators did was lower the bar to creating images that were believable.”
Carson was also interested in how these fakes were made with AI — and whether the process of AI “learning” was, in fact, a form of copyright theft. To test the hypothesis, he used prompts to direct an AI to produce an image of a protester wearing an American flag T-shirt, throwing a tear gas canister during the Ferguson Uprising.
Those elements could be presented in an infinite number of varieties. But Carson was stunned watching his prompts quickly return an image nearly identical to the photo taken by Post-Dispatch photojournalist Robert Cohen on Aug. 13, 2014. The image was part of the 2015 submission that won Carson and Cohen a Pulitzer Prize for their coverage of the summer protests in 2014.
To Carson, the fact that the AI had apparently resorted to copying Cohen’s image indicates that the technology isn’t really “learning” when it processes training data or “scrapes” images from public websites — it’s stealing.
Police fire tear gas at protestors for the third night in a row in Ferguson, Mo. #Ferguson #MikeBrown pic.twitter.com/BEoZT5yjkU
— Robert Cohen (@kodacohen) August 13, 2014
“I thought that Robert’s photo would be a good example, because it is an iconic photograph, probably the best-known photograph from the protest,” Carson said, explaining his experiment. “Within a few pretty simple prompts, we ended up at something that I think was pretty clearly a copyright violation. That was really troubling to me.”
He continued: “I think it confuses the public as to what's real and what's not. We're used to trusting our eyes. And I'm sort of fascinated with us being in this world, in this time now, where it becomes more difficult to trust what we see.”
To hear the full interview with photojournalist David Carson, including insight from his research with AI images, as well as his argument that AI companies have a copyright problem, listen to St. Louis on the Air on Apple Podcasts, Spotify or YouTube, or click the play button below.
“St. Louis on the Air” brings you the stories of St. Louis and the people who live, work and create in our region. The show is produced by Miya Norfleet, Emily Woodbury, Danny Wicentowski, Elaine Cha and Alex Heuer. The audio engineer is Aaron Doerr. Send questions and comments about this story to talk@stlpr.org.
