40-year-old Adam Johnson spent hours researching the ballot on the porch of his Morgantown, West Virginia, home — but not with a relative or neighbor. He used that time to hash out his midterm choices with ChatGPT.
Johnson is part of a group of voters turning to AI chatbots for election homework, with the tools proving especially appealing in lower-profile contests that receive less coverage.
Here's what to know
According to NPR, Johnson — a graphic designer training as an illustrator — had ChatGPT list every contest on his ballot, then used it to weigh each candidate against his own priorities. He said he usually backs Republicans unless a contender is tied to the Make America Great Again movement, and he asked the chatbot to work through topics such as immigration and the Senate contest between Shelley Moore Capito and Rachel Fetty Anderson.
Citing the Pew Research Center, NPR reported that roughly half of adults in the United States have used AI-powered chatbots and that 42% of chatbot users say they use them to find information. A September survey from the American Association of Political Consultants Foundation found that about one in five voters use AI chatbots for election news and voting.
Rafael Batista, a Johns Hopkins University postdoctoral fellow who researches AI's effects on society, said these tools' appeal can come with a downside: they may strengthen views users already had.
"It might select some things that would reinforce and persuade you even more towards the way that you were leaning already," he told NPR. "So you leave more confident, without necessarily learning more about the world."
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
"ChatGPT is just kind of a people pleaser," he said. "Whatever you lean towards, it will speak positively about it."
More background
NPR also heard from voters using chatbots in other ways. In Wisconsin, retired Madison City Council staffer Lisa Veldran turned to Gemini after Sara Rodriguez exited the governor's primary and used it to generate a side-by-side table comparing Francesca Hong and David Crowley.
Exton, Pennsylvania, author Lakshmi Iyer, 50, used a chatbot for something else entirely: building Ballot Lookup with Claude. The tool, described as an "independent, volunteer-built, nonpartisan lookup," shows Pennsylvania voters who appears on their ballot and what positions candidates have taken. She said she tried to limit bias by directing the chatbot to rely on candidates' speeches and official websites.
That trend is unfolding as many communities have less local political coverage than they once did. Tyler Black, a Nashville, Tennessee, software development executive, said chatbots help him research local contests, though he cross-checks one bot against another because fabricated or mistaken answers remain a risk.
What can be done?
For voters using chatbots to study their ballots, Batista recommended writing prompts that reveal as little as possible about their own politics. He also said incognito mode may help by reducing how much prior information the chatbot can draw on.
Black compares answers across multiple chatbots instead of trusting a single output, while Iyer tries to anchor results in original campaign materials rather than broad summaries that may reflect the model's assumptions.
An OpenAI spokesperson referred NPR to the company's election information and safeguards page, which says it continues to "monitor bias in our models to keep ChatGPT's responses politically neutral."
Even with those steps, Veldran said the tools should remain in perspective.
"You also have to keep in mind what's running these chatbots and AI in general," she said. "Who's paying to build these bots and what is their ultimate goal."
Johnson said county commission races are still up for review, but he already feels the chatbot has prepared him better.
"I wouldn't feel too bad if I had to do the election today," he said.
Where can I learn more?
Questions about AI influence, accuracy, and public trust extend well beyond ballot research, as this related reporting shows.
• Generative AI systems are amplifying harmful patterns as misinformation spreads online.
• BBC testing showed major AI chatbots were seriously flawed at summarizing the news.
• In London, researchers are seeing more AI chatbots ignoring instructions, raising harm concerns.
Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.
