This is a Civics 101 story from CNY Decides, a collaborative podcast between the award winning newsrooms of WAER and Central Current.
As the primary election approaches on June 23rd, it’s tempting to search online for candidate races and other information.
In today's digital world, voters are increasingly relying on artificial intelligence to fact check. Afterall, it takes only minutes to hop on a computer and enter a few words to yield what looks like solid answers.
It is easier, say experts, but they warn it is often misleading. In Los Angeles, for instance, a fan of mayoral candidate Spencer Pratt used AI to create a series of superhero video clips showing him dressed as Batman defeating Hollywood personalities and politicians. It was absurd, but impressively imaginative.
“Most political advertising is misleading to a certain extent because you’re trying to bring someone over to your side,” Syracuse University Professor Beth Eagan flatly admitted.
The advertising professional started working on political campaigns in the 1980s, when a 30 second commercial took an entire crew.
“If we wanted to get an ad out. It was weeks of development and filming,” whereas today, she said, “anyone can do it within seconds.”
In the 80’s, up until recently, political ads ran mostly on traditional TV, radio, and billboards, where words and images are held accountable by federal communication and election laws.
The marriage of AI and social media allows political messaging to bypass regulators, while looking and sounding real, and worse, feeding trapped mindsets.
“We love the echo chamber,” said Edward Watson, “We'd love to hear people tell us they have the same perspectives and that our ideas are correct and how smart we are as a result.”
Watson heads up digital innovation at the American Association of Colleges University.
The beauty and battle of depending on AI is its built in hallucination effect.
“It’s designed to be creative and to think outside of the box and to give us at least the impression of new information,” Watson confirmed, “So as a result of that, it often will essentially make up things.”
It works based on the individual operator who puts in the query and the rules to follow. Ultimately, the system scrapes the internet for information it calculates you will want to hear.
Outside the user’s sphere, “It doesn’t understand context,” Eagan noted. “It is simply doing mathematical equations to calculate the probability that word or that image, that pixel, is the next best thing.”
To break that bubble, there are a number of ways to safeguard your decisions before going into the voting booth.
First, see where the news source you currently trust sits on a bias and truth chart tracked by Ad Fontes Media, a Colorado based media watchdog agency. If it’s not in the upper center region, pick three that are, preferably on both sides of the bias axis.
When searching online, do so using a privacy option like Google Chrome’s Incognito mode or DuckDuckGo’s private mode.
And, wherever an AI summary offers a link, opt instead to find the info it has harvested directly at the source’s own website to see statements and findings firsthand.
Look for resources that rely on scholarly studies such as Google Scholar, a platform offering literature, articles, theses, books, and abstracts published by academic scholars.
Academic databases like JSTOR, IEEE XPlore, PubMed, and university libraries offer peer-reviewed articles and research papers.
The goal is to triangulate the resources to the point where there are three independent and neutral sources that have come to the same fact-based conclusion without bias or opinion.
