An image purporting to show a young survivor of Hurricane Helen appears to have been created using artificial intelligence tools. The image was viewed millions of times online, even after its origin was questioned. hide caption
toggle caption
After images of the devastation left by Hurricane Helen began circulating online, so too did images of a crying child holding a puppy on a boat. Some of the posts on X (formerly Twitter) that featured the image received millions of views.
This prompted an emotional response from many users, including many Republicans who wanted to criticize the Biden administration’s disaster response. But others were quick to point out telltale signs that the image was likely created with a generative artificial intelligence tool, such as limb malformations and blurring that are common with some AI image generation tools.
This election cycle, such AI-generated composite images have proliferated on social media platforms, often following political news events. Online platforms and people watching the election closely say these images are a way to spread a partisan narrative where the facts are often irrelevant.
According to Rolling Stone, after X users added a community note warning that the image of a child on a boat was likely generated by AI, Sen. Mike Lee (R-Utah) and others, Some people who shared the image deleted their posts about it. .
But others doubled down even after the source of the image composite was revealed. “I have no idea where this photo came from, and honestly, it doesn’t matter,” Amy Kremer, a Republican National Committee member representing Georgia, wrote of X.
“It’s a form of political propaganda, a way to show interest in and support for a candidate in a fandom-like style,” said Renee DiResta, a professor at Georgetown University’s McCourt School of Public Policy, who recently wrote: I wrote a paper. A book about online influencers. “Political movements can then pick it up, retweet it, and promote it, so that they’re seen as participating in the conversation, or as if they were participating in the joke itself.”
Other images that appear to have been created by AI also circulated after hurricanes Helen and Milton, depicting animals on rooftops just above flood waters. After former President Trump and vice presidential candidate J.D. Vance amplified baseless claims that Haitian immigrants in Springfield, Ohio, were eating pets and wild animals, AI-generated President Trump Images of people hugging each other flooded X and other social media platforms popular with Republicans.
Generative AI is another tool for supporters to interact with campaigns online, DiResta said. “It’s cheap, easy, and fun, so why not give it a try?”
truth and facts in images
In the same post defending her decision to continue posting the composite image, Kramer also wrote, “This is symbolic of the trauma and pain that people are living with.”
Matthew Burnidge, a professor at the University of Alabama who studies online news deserts and political communication, says the separation between facts and deeper ideas of truth also has implications for Western philosophy. “If you go back and dig into the works of Kant, Kierkegaard, and Hegel, the idea that there is some kind of deeper truth that is often associated with freedom or the sublime or something along those lines comes up. ”.
To be clear, research suggests that individual fact-checks against politicians can add up and change how voters feel about them. One study shows that fact-checking changes Australians’ feelings about politicians. But another study showed that fact-checking Trump did not change Americans’ views about him even as their beliefs about individual facts changed.
Emily Vraga, a health communication researcher at the University of Minnesota, said fact-checking images can be more difficult than text. “There’s a lot of research showing that people have a very hard time knowing what’s real and what’s not when it comes to online images,” Vraga said. “This has been true even before ChatGPT. ” he said.
Capturing visual images can evoke emotions before people can process what they’re seeing. A team of researchers examined Pinterest posts depicting vaccine needles that included misleading images containing oversized needles and brightly colored liquid.
“The needle is much smaller. This is not a very convincing correction,” Vraga said. “It’s part of a larger narrative that vaccines are unnatural and dangerous.”
The surreal and often creepy images generated by AI can exist in a gray space between fact and fiction for viewers. A photorealistic image of pop star Taylor Swift endorsing Trump is clearly not Swift upon closer inspection, but the fleeting resemblance makes it clear, said New York University art historian Ara Merjan. He said it had an impact on those who saw it. “If someone had drawn Taylor Swift in a comic supporting Trump, it wouldn’t have been a scandal.”
The pop star cited AI-generated imagery and “misinformation” as one of the reasons she supports Vice President Kamala Harris as president.
Driven by money, AI slop overwhelms online space
AI imagery has also partially filled the space left by traditional news media as the news industry shrinks and technology platforms deprioritize news. “Who’s moving into that space? The propagandists,” Burnridge said. “Propaganda often penetrates in other ways, not as news, but through lifestyle content.”
Politically influenced images are just a small portion of the images generated by AI on social media platforms. Researchers discovered an AI-generated cake, a kitchen, and a shrimp-like Jesus rising out of the sea. Some competed for website traffic, while others tried to steal personal information or money from viewers.
People in developing countries are teaching others how to use AI-generated images to create trending posts as Facebook pays them to create popular content, according to a study by 404 Media. It turned out that. Payments can be higher than the typical local monthly income. Many of the images produced by these content farms evoked strong, sometimes patriotic feelings. Some images looked realistic, while others were more artistic.
An AI-generated image purporting to show Vice President Kamala Harris in communist costume. The image was shared by X owner Elon Musk to his millions of social media followers in August. hide caption
toggle caption
Danger to elections
One of the most memorable AI-generated images related to politics was backed by X owner Elon Musk. It depicted a person resembling Harris, wearing a red military uniform with a hammer and sickle on his hat.
Eddie Perez, a former Twitter employee who focuses on election confidence at the nonpartisan nonprofit OSET Institute, said the image was meant to portray Harris as un-American. Ta. “The message is that there is no legitimate way for Kamala Harris and her party to actually win the presidential election.”
Perez said such images foster political polarization and can undermine people’s confidence in the election results. Republicans have suggested for months that Democrats are likely to steal the election from President Trump through a variety of subterfuges.
“There are so many different ways and modalities for that strategy to be implemented. Generative AI is just one tool in the toolkit, so to speak. Or do you think it’s worse than many other tools? No. Does that mean it’s benign? No,” Perez said.
