Taylor Swift and AI
NPR: STEVE INSKEEP, HOST:
Artificial intelligence is playing a role in the presidential campaign. Last week, for example, former President Trump posted an AI-generated image of Taylor Swift. He thanked her for the endorsement, which she never gave. She's made no endorsement this year and, in 2020, backed President Biden. NPR's Shannon Bond assesses the real-world effects of fake images.
SHANNON BOND, BYLINE: The depiction of Taylor Swift Trump posted on Truth Social is obviously fake. So were some of the other images he shared with it of young women in T-shirts reading Swifties for Trump, and Trump himself knows that. Here's what he told Fox Business a few days late
DONALD TRUMP: “I don't know anything about them other than somebody else generated them. I didn't generate them. Somebody came out. They said, oh, look at this. These were all made up by other people.”
Excuse me, but if you believe that I want to sell you land in Antarctica.
SHANNON BOND: He even warned about the dangers of artificial intelligence showing people doing or saying things they didn't.
DONALD TRUMP: “It's happening with me, too. They're making - they're having me speak. I speak perfectly, I mean, absolutely perfectly on AI. And I'm, like, endorsing other products and things. It's a little bit dangerous out there.”
Yeah, and he is one of the biggest perpetrators of that warning.
SHANNON BOND: But in reality, Trump himself has been sharing AI content on social media, often made by his supporters who have embraced new tools that make it easy to turn quick text descriptions into images, both plausible and fantastic. The Trump campaign didn't respond to a request for comment. Frequently, the subject of the images fans create is Trump himself - astride a lion, as a bodybuilder, dancing with Elon Musk. Other times, it's his opponents, like a recent fake image he shared of what appeared to be Vice President Kamala Harris addressing a Soviet-style rally, complete with a Communist hammer-and-sickle flag.
BRENDAN NYHAN: One thing AI, in particular, is good at doing is taking ideas or claims being made in the text and turning them into symbolic representations.
BOND: Brendan Nyhan is a political science professor at Dartmouth College. He says this is an AI-powered spin on the memes, satire, and comedy that are a staple of political communications.
BRENDAN NYHAN: It's chum for the folks who read political news online and tend to have slanted information diets.
SHANNON BOND: These AI images are often shared as jokes or parodies. The intent is not for them to be taken as literally true. Some of the fake Swifties for Trump pictures Trump posted were labeled as satire by their original creator. As AI tools have become widely available, worries have focused on their most malicious uses to deceive voters, like deepfakes of candidates or photorealistic depictions of events that never happened. But more mundane uses of AI can also contribute to an environment where people trust what they see less and less. Trump himself has capitalized on that distrust, claiming falsely that AI faked a real photo of a Harris campaign rally. Hany Farid is a professor at UC Berkeley who specializes in image forensics.
HANY FARID: If you constantly see fake images and ones that look really compelling and really highly photorealistic, when you see the large crowd, are you then less likely to believe it? Do you now start just to doubt everything?
SHANNON BOND: There's still plenty of time before Election Day for deceptive or misleading AI content to go viral - or, as some observers say is the greater risk, to be targeted at individual voters in private channels. Daniel Weiner is director of the Elections and Government Program at the Brennan Center for Justice, a public policy nonprofit focused on voting rights. He's worried about AI fakes that might discourage people from voting, like false claims of threats or outages at polling places.
DANIEL WEINER: The biggest concern around AI is that it tends to amplify threats that result from other gaps and weaknesses in the democratic process.
SHANNON BOND: He fears the technology could be used to erode trust in elections themselves further.
Shannon Bond, NPR News.
Copyright © 2024 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.
An NPR contractor creates NPR transcripts on a rush deadline. This text may not be in its final form and may be updated or revised. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record
Comments