AI Chatbots Swaying Voters: Beyond Political Ads

The Use of AI Chatbots in Political Campaigns

In 2024, a Democratic congressional candidate in Pennsylvania, Shamaine Daniels, used an AI chatbot named Ashley to engage with voters. The calls began with, “Hello. My name is Ashley, and I'm an artificial intelligence volunteer for Shamaine Daniels’s run for Congress.” Although Daniels did not win, preliminary research suggests that these AI-driven conversations may have positively influenced her campaign.

AI's Impact on Voter Opinion

A multi-university research team found that conversations with a politically biased AI model could effectively sway voters’ opinions, often more so than traditional political advertisements. The AI chatbots influenced opinions by presenting facts and evidence, although the accuracy of these claims varied significantly.

Persuasive Findings from Recent Studies

The findings were published in Nature and Science, showcasing AI's evolving persuasive capabilities. Research suggests that a single conversation can significantly affect voter choices. Notably, Donald Trump supporters showed a 3.9-point shift toward Kamala Harris after interacting with an AI favoring her, compared to the effectiveness of political ads in previous elections.

Mechanisms of Persuasion

Gordon Pennycook, a psychologist involved in the Nature study, notes that LLMs generate real-time information, allowing for strategic conversation flows. In studies leading to the 2025 elections in Canada and Poland, the chatbots managed even larger shifts in voter attitudes—up to 10 points.

The Concerning Accuracy of AI-Generated Information

Despite their effectiveness, many of the facts presented by AI were incorrect. Research showed that chatbots promoting right-leaning candidates propagated more inaccuracies than those on the left. The underlying LLMs draw from extensive human-generated text, reflecting real-world biases in political communication.

Optimizing Persuasiveness vs. Truthfulness

Another study demonstrated that the most convincing chatbots employed strong facts and were trained with examples of effective conversations, achieving a 26.1-point shift in polarized participants. However, this pursuit of persuasiveness came at the cost of truthfulness, as accuracy diminished when attempts to persuade increased.

Implications for Democracy and Future Campaigns

The authors argue that AI-powered campaigning could significantly shape voter perceptions, possibly compromising independent judgment. The potential for misinformation to spread further complicates the issue, with concerns about the balance of truth and falsity in political discourse. Andy Guess raises critical questions about how AI might evolve in the campaign landscape.

The Path Forward: Regulation and Ethics

As AI chatbots gain traction, voters may start relying on them for electoral guidance. The perpetual need for strategies that ensure accuracy and accountability in AI is paramount to safeguarding democratic processes. Establishing rigorous auditing practices for LLM outputs in political contexts could serve as crucial initial measures.