Early in Poynter’s community conversation Tuesday night, a visual was displayed for the audience that sparked laughter, amusement and confusion. Here was Jesus Christ, underwater. His eyes were closed, and his crown of thorns appeared behind him like a halo. But his outstretched arms were not, in fact, arms. They had been transformed into beady-eyed shrimp. His body was also covered in the crustaceans.
“This is central to the online experience now,” said Alex Mahadevan, director of MediaWise and a member of Poynter’s faculty. “This is an AI-generated image that was created to basically garner a ton of engagement, which it did. This is just one example of ‘Shrimp Jesus.’”
An image from a presentation from Poynter’s community conversation, The Future of Facts Online. The images in this screenshot were generated by artificial intelligence and do not reflect reality. (Screenshot/Poynter)
Mahadevan, an expert in AI, pointed out that artificial intelligence is creeping into every corner of online life, from Google searches to our social media feeds. The problem, he cautioned, is that visuals generated with AI also appear during disasters. He referenced an image of a child in a rescue boat. The girl wearing an orange life jacket — purported to be a young victim of Hurricane Helene — was crying while holding a puppy. Around them, there appeared to be major flooding.
“You may have seen this image of the crying child that went viral. It looks hyperrealistic,” Mahadevan said. “It was used politically.”
An image from a presentation from Poynter’s community conversation, The Future of Facts Online. The images in this screenshot were generated by artificial intelligence and do not reflect reality. (Screenshot/Poynter)
Aaron Sharockman, Poynter’s vice president for sales and strategic partnerships and moderator of the panel event, asked the audience how many had seen this image. “I saw it on my feed from a former journalist, actually, who shared it. And people said, ‘That’s AI,’” he recalled. “And then their response was, ‘Well, whatever. I care about the people in North Carolina.’ That’s kind of where we are.”
Mahadevan stressed that this is the heart of the issue with AI’s omnipresence — a willingness to believe something is true if it conforms to your viewpoints.
“It doesn’t matter anymore,” he said. “If what you see online confirms how you feel about the world and what you believe, and whether the good guy is right and the bad guy’s wrong, then it doesn’t matter.”
Bleakness around the current rise of AI-driven content, challenges to online trust, and the dismantling of fact-checking by social media platforms were the major focuses of the event held Tuesday night at the Poynter Institute in St. Petersburg. They also raised the alarm on the implications for society and young people. Sharockman led the panel discussion between Mahadevan; Drew Harwell, a technology reporter for The Washington Post; and PolitiFact editor-in-chief Katie Sanders.
Aaron Sharockman, left, Poynter’s vice president for sales and strategic partnerships, moderates a panel featuring, left to right, Drew Harwell, a technology reporter for The Washington Post; Alex Mahadevan, director of MediaWise and a member of Poynter’s faculty; and Katie Sanders, editor-in-chief of PolitiFact at the Poynter Institute in St. Petersburg, Florida, on Tuesday, May 6, 2025. (Chris Kozlowski/Poynter)
In response to Mark Zuckerberg’s vision of AI chatbots becoming a part of one’s friend network, Harwell said he found the idea absurd but the kernel of it to be interesting “because these chatbots are becoming more important to our lives.” He referenced Gemini, a generative AI chatbot developed by Google, and how there were plans to roll out Gemini to child accounts. Now, adults would need to have a conversation with children to explain that Gemini is not real.
“It was a little unnerving to me to think about replacing people, because you can’t replace people in that way. But, also, these are the tech companies that we have to deal with,” he said. “This is the reality that they’re sort of imposing on our society. So it’s good for us to be aware of how these companies are operating and understand the limitations in these chatbots. They’re not people.”
Though there are concerns around chatbots, Mahadevan pointed out that some newsrooms are using these tools for good. He shouted out Haystacker, a proprietary tool from The Washington Post that allows its journalists to sift through large amounts of data.
On chatbots, Mahadevan said a good friend will tell you if you have something in your teeth. “These chatbots are not designed to do that. They are designed to tell you everything that you want to hear,” he said. “And looking at the issues in the world, I could say being told everything you want to hear and that you are the best is probably at the heart of a lot of issues we have. So this is central to our online experience now. It’s scary.”
The panelists also spoke at length about the Trump administration’s use of AI. The White House has hopped on a trend that generates AI images mimicking the style of Studio Ghibli, a well-known Japanese animation studio. It shared an image of a Ghibli-style meme of Virginia Basora-Gonzalez, a woman who had been deported from the U.S. several times and was convicted of fentanyl trafficking. In the image, the woman wears a long-sleeved black shirt and grey pants, and her brown hair is wrapped under a hairnet. Big teardrops fall from her face. The real Basora-Gonzalez cried during her arrest, and the actual photos of her were also posted by the White House on X.
An image from a presentation from Poynter’s community conversation, The Future of Facts Online. The images in this screenshot were generated by artificial intelligence and do not reflect reality. (Screenshot/Poynter)
The style of Studio Ghibli’s work is often seen as cozy and warm.
“It was a juxtaposition of this very light-hearted animation thing, with this political messaging. And when you see that, it did strike me as pretty cruel, just on the human level. But if you are supportive of that immigration status, you’re going to like that because you get the joke,” Harwell said. “And so what I find really interesting about this is you’re not going to believe that that was something that somebody maybe drew by hand. But you see the message.”
President Donald Trump also recently shared an AI-generated photo of himself dressed as the pope on Truth Social, then later said he had nothing to do with it. Harwell noted that the White House has a whole digital strategy team — a focus of some of the Post’s journalism.
“They want to be seen in whatever way possible,” he said. “And AI is great for that, because it takes five seconds to put out a photo like that. It’s very evocative.”
AI joins a long list of sources of forgeries. Sanders said a lot of old tricks, like simple Photoshops, still work really well for people in power. She brought up a recent fact check from PolitiFact about an image of the tattooed hand of Kilmar Armando Abrego Garcia, a man whom the government deported to El Salvador two months ago despite a court order not to deport him to the Central American country. Trump had posted an image of what he said was Abrego Garcia’s hand. To Trump and his administration, the images were proof that he was a member of MS-13, the international criminal gang.
“But then he was taking the thing that was created in Microsoft Paint — basically the oldest program there is — to say that his hand literally said, ‘M-S-one-three,’” Sanders said. “Just because he’s saying it creates enough doubt in our society, that this photo is also real because he’s saying it is. And so it can be complicated, or completely made up. And it’s also quite easy.”
Sharockman pointed out that though Sanders and her team fact-checked the circumstances around the image of Abrego Garcia’s hand, they didn’t do the same with the photo of Trump in papal robes. In a moment of levity, Sanders said they didn’t even think to do so, causing the crowd to erupt in laughter.
Outside of the White House, Sharockman asked what could be gained from people using AI across social media platforms. He questioned what makes them want to push false narratives. Harwell said there’s a political clout to be gained. He said many people now use social media as a passive news source. They just scroll.
“The currency of social media is attention,” he said. “All of these people want attention. Attention is how you get followers. It’s how you get brand deals. It’s how you get famous. It’s how you build the life you want to build. And so for some people, they do that by doing it the right way — sharing legitimate information, doing journalism, reporting. Some people just post whatever — AI BS. And so it can get really hard when you’re scrolling to use that media literacy and understand what you’re looking at.”
Tuesday’s conversation also turned to Meta ending its partnership with independent American journalists, including PolitiFact, which served to pinpoint and call out hoaxes or other misleading information on its platforms. Sanders said it feels like we lost a “really helpful tool” to help people figure out what is real and what’s not on platforms like Facebook, Instagram and Threads.
“I’m personally still a little sad,” she said. “Not just for the business loss, but for the effectiveness of the partnership in making those platforms more consumer-friendly, to be honest.”
Mahadevan described the replacement of professional journalists and fact-checkers with the general public as a monumental failure. “The future of facts online is you,” he told the audience. “In an incredibly hostile online world, all of these platforms have basically said, ‘You’re on your own. It’s up to you.’”
During the event’s question and answer portion, a father spoke about his concerns about his children in this age of social media and widespread misinformation. If the panelists had a magic wand, he asked, would they go back and delete social media from existence? Harwell acknowledged feeling very cynical some days, but opted to be naive for a second. He described seeing angry people on social media for much of his day.
“When I talk to some of those angry people in person, it’s amazing how much we can actually find to agree on,” he said. “And so when I see young people on social media, what really excites me is that they are still engaged to learn.”
He said they are scrolling through their feeds because they want to learn. They don’t want to be lied to.
“They do want to know the facts, just like a lot of us in this room,” he said. “So I think maybe we’ll start seeing more in-person communities, news communities where we actually engage with people as opposed to just looking, tapping at screens all day.”
Comments