Albany, NY — As artificial intelligence (AI) takes a more prominent role in political campaigns, a new study from the University at Albany highlights how voters’ trust can vary based on their political affiliations. Researcher Seulki Lee-Geiller examined this phenomenon, revealing that perceptions of election integrity are influenced not just by AI itself but by the political party believed to be utilizing it.
The study, titled “AI as a partisan cue: the relational dynamics of AI-assisted campaign messaging in election integrity perceptions,” was published on September 24 in the Journal of Elections, Public Opinion and Parties. It analyzes data from an online survey of 3,055 U.S. participants conducted in June and July 2024, just before the presidential election.
In the study, participants were randomly divided into four groups and showed a hypothetical campaign letter from a presidential candidate. While the content remained unchanged for all participants, variations included whether AI had a role in drafting the letter, and if so, whether it was employed by their favored party or the opposition.
After reviewing the letter, participants evaluated several aspects of election integrity, including fairness, authenticity, freedom from fraud, and representation of voter interests. Lee-Geiller noted that awareness of AI’s involvement did not significantly alter perceptions of the letter’s fairness and authenticity. However, respondents did express concerns about targeted messaging using AI.
Significant differences appeared when political affiliation was factored in. When participants believed their preferred party used AI, they rated election integrity more favorably than when they thought the opposing party was employing the technology. On a five-point scale, the average authenticity rating was 3.71 when AI was linked to the preferred party, compared to 3.37 when linked to the opposing party.
Lee-Geiller emphasized the findings’ implications, stating, “As AI becomes more common in political communication, we need to think beyond what the technology can do and consider how people understand and respond to it. Transparency, consistent standards, and digital literacy can help people make more informed judgments about AI-assisted messages, regardless of who is using the technology.”
The research is described as exploratory and focuses on immediate reactions to hypothetical campaign materials rather than real-world electoral scenarios. Lee-Geiller mentioned the need for further investigation into how perceptions may change with ongoing exposure to AI-driven political messaging in actual campaigns.
