Ricardo Martinez-Guerrero, a political science and history major at San Diego City College, thinks that, like news organizations that lean left or right, social media pushes out content or customizes it in the same way.
“There’s people (who) resort to the New York Times, Fox News, CBS, ABC, etc., for life, because that’s reassuring to their narrative (of) what they already think,” Martinez-Guerrero said. “I think it’s very similar to what AI does.”
In an article posted by Science, researchers experimented with the idea of political persuasion with conversational AI.
They found that when AI systems are used for persuasion, they are trained to put out factual claims until they run out of points, asserting the idea that the system will increasingly generate misleading information.
AI chatbots use large language models (or LLMs for short), which are advanced systems that generate human-like text by understanding and processing massive datasets that translate, summarize, and create content distributed through social media.
Jeanette Lowrie, a cybersecurity major at City, explained how influence campaigns on social media could be run by a network of robots.
These bot campaigns share repetitive content and have predictable patterns, making it easy to detect when one of these accounts on social media doesn’t have an actual person behind it.
“So, (when) it’s related to elections, bots try to make posts that influence people’s opinions about candidates and other political matters,” Lowrie said.
This has become a concern, as it could be used for political persuasion, especially if many college students are receiving their news from social media platforms, according to Tufts Circle.
Nearly 75 percent of students find most of their news through social media, according to the Family Online Safety Institute.
News can be altered about candidates without students really giving them a fair shot at showing politicians, if you’re assuming based on one headline, according to Tufts Circle.
It is not enough when social media platforms fail to catch problematic content before it’s posted or regulate certain content on AI platforms, according to Brandi Wampler in his research at Notre Dame News.
“(Social media sites) Reddit, Mastodon and X were trivial,” Wampler wrote. “Despite what their policy says or the technical bot mechanisms they have, it was very easy to get a bot up and working on X. They aren’t effectively enforcing their policies.”
Social media platforms have incorporated AI into their platforms, so when a user swipes on their timeline or explore page, AI is controlling the content recommended and the ranking the user receives.
“We’re at the misinformation age,” Martinez-Guerrero said. “So I think, especially with the complicity that AI companies have with the current administration, there is, undeniably, there is bias with AI.”
More than half of U.S. states have laws regulating deepfakes with political messaging.
Lowrie echoed Martinez-Guerrero, saying social media companies need to be stricter when it comes to holding bots accountable.
“Companies could do more to identify and not give algorithmic preference to like bot accounts and stuff,” Lowrie said.
Martinez-Guerrero explained that people seek to assert themselves based on what they see on social media. With social media integrating AI and a content recommendation system, it only affirms the ideas.
“I think that has been a topic with social media algorithms and in general, with news channels,” Martinez-Guerrero said. “People love seeking reaffirming evidence for what they already believe.”
Questioning the content shown to you by social media and AI is necessary in a time when everything is curated for you, Lowrie said.
“You just have to be willing to question divisive content when you see something that clearly has some kind of very easily digestible message,” Lowrie said. “Question,’what is that message? And why is it being served to you?’”
While using generative AI might be easier to use, it could sway a student’s opinions without seeking evidence or sources.
Students’ best option is to research and to find multiple sources, not just assume what they’re reading is factual, according to Standford Report.
“(Seek out) some real local journalists producing voting guides (that) don’t tell you how to vote,” Lowrie said. “They’ll explain the measures and the candidates and things like that to you, and then you can make up your own mind.”
City Times Media will be releasing its own voting guides. Look for that in the election section of sdcitytimes.com.
This story was edited by David J. Bohnet and Itzel Martinez.
