With more than 70 countries hosting national elections, 2024 is the biggest election year in history, according to The Economist.
But how misinformation impacts elections, especially with the rise in content generated by artificial intelligence, continues to be of concern.
A research team examined misinformation narratives on social media in 2023 regarding the Taiwanese presidential election on January 13, 2024. They were especially interested in how narratives targeted relations between Taiwan and the United States.
Misinformation targeted mistrust and skepticism toward the U.S. rather than U.S. foreign policy directly, according to the study, which was published in the Harvard Kennedy School Misinformation Review.
To examine misinformation across three platforms: Line—the most popular messaging app in Taiwan; PTT—the Taiwanese equivalent of Reddit; and Facebook, the researchers applied large language models—AI algorithms like ChatGPT that can understand and generate text and other content.
When Line users come across content that may be dubious, they have the option to forward it to the Cofacts chatbot, a third-party, crowd-sourced platform in which content is fact-checked by volunteers.
Of the 140,300 Line posts relating to the Taiwanese election, which were fact-checked by Cofacts and analyzed by the researchers, nearly 11,000 posts focused on issues relating to Taiwan, the U.S., and China, making it the most prominent fact-checked topic.
The study found that misinformation toward the U.S. contained various narratives, such as the U.S. and the threat of war, especially regarding Ukraine; economic decline due to fiscal actions by the U.S.; and the vaccine supply and the U.S.
For example, some posts suggested that the U.S. deliberately limited the COVID-19 vaccine supply during the pandemic while others stated that U.S. vaccine doses were cheaper domestically than what Taiwan had to pay.
This misinformation may have been planted on social media to damage the United States’ reputation and credibility and “lower the Taiwanese audience’s belief that the U.S. will help,” according to the researchers, as prior research has provided evidence of such actions.
The team found that much of the misinformation on U.S.-Taiwan relations circulated most prominently within China-friendly political groups and referenced U.S.-based conspiracy groups.
“More than half of the misinformation content on the U.S. relating to Taiwan’s recent presidential election in our datasets was multimodal,” says lead author Herbert Chang ’18, an assistant professor of quantitative social science at Dartmouth. “With so much visual content, our results contribute to the literature and concerns about the impact of AI-generated content on elections.”
By tracking misinformation activity and the location of where the content was posted, the data showed that some users in China and other areas in Asia deliberately masked their geolocations by using virtual private networks to make it look like they were based in the U.S., when in fact that was not the case.
Taiwan’s politics have typically been marked by two parties, the Nationalist Party led by Kuomintang, or KMT, and known as the blue camp, which reflects a more pro-integration sentiment towards China, and the Democratic Progressive Party, known as the green camp, which has wanted to change the country’s name from the Republic of China to Taiwan and modify the constitution. A third party, the Taiwan’s People’s Party, emerged in 2020 with a call for greater attention to domestic and social issues and less emphasis on foreign relations. The analysis builds on research studying social media discourse and support for the presidential candidates, published in PNAS Nexus earlier this year.
Lai Ching-te of the DPP was inaugurated on May 20 as Taiwan’s president, as the DPP won the election for a third consecutive presidency with 40% of the votes. The KMT received 33% of the votes and the TPP received 26% with strong support among 20-somethings.
“This election not only demonstrates the increased importance of international relations in voter choice around the world, at a time of heightened geopolitical conflict, but also how AI can be used to identify and fight misinformation, not just create it,” says Chang.
Austin Wang at the University of Nevada, Las Vegas, and Yu (Sunny) Fang at Barnard College also served as co-authors of the study.
