Early voting is already underway in the 2026 midterm elections, and a growing number of Americans are adding an unusual new tool to their civic routine: artificial intelligence chatbots.
Rather than relying solely on traditional sources such as newspapers, candidate forums, or conversations with friends and neighbors, some voters are now asking large language models like ChatGPT to help them research their ballots, fact-check political claims, and compare candidates on the issues.
The trend is highlighted in a recent NPR report that profiled voters who have folded AI tools into their decision-making process. Among them is Adam Johnson, a 40-year-old graphic designer from Morgantown, West Virginia, who is studying to become an illustrator.
On a September evening, Johnson sat on his porch with his ballot, working through the various races and candidates to determine who best matched his values. Instead of talking it over with a friend or relative, he spent hours in conversation with ChatGPT, the chatbot developed by OpenAI, a prominent artificial intelligence research company.
For Johnson, the technology is already woven into daily life. He said he uses ChatGPT for meal planning, help with his job search, and a range of other household tasks, making it a natural place to turn when election season arrived.
His approach reflects a broader shift in how people interact with information. AI chatbots can process enormous amounts of publicly available text and respond to follow-up questions in plain language, offering a level of convenience that a search engine or a stack of voter guides may not match.
That convenience, however, raises questions about accuracy, bias, and the limits of machine-generated advice. The source article notes that when Johnson had raised political topics with ChatGPT in the past, the system initially appeared reluctant to engage, giving a simple refusal rather than offering an opinion. That reluctance is a common feature of many commercial chatbots, which are often programmed with guardrails intended to avoid overtly partisan or misleading statements.
Yet Johnson's experience suggests those guardrails are not absolute. With the right prompting, and with the user guiding the conversation, the chatbot did engage with his ballot questions, helping him work through the candidates and races in a back-and-forth discussion lasting hours.
The NPR report frames this as part of a wider pattern. Voters are increasingly using AI chatbots both to verify factual claims made by campaigns and to weigh competing candidates against their own priorities. Rather than replacing human judgment, the tools appear to be functioning as a kind of interactive research assistant for people who want to dig deeper before casting a ballot.
The timing is notable. Midterm elections typically draw lower turnout than presidential contests, and voters often report feeling underinformed about down-ballot races, where local and state offices receive far less media coverage. In that environment, an AI tool that can generate a summary of every race on a ballot may hold real appeal.
Chatbots also offer a level of privacy that some voters may prefer. Asking an AI about political preferences avoids the potential awkwardness of discussing politics with friends, family, or coworkers, and it allows users to explore questions at their own pace, at any hour, without fear of judgment.
Still, the practice is not without risk. Large language models are known to occasionally produce inaccurate or fabricated information, a phenomenon often described as hallucination. When the subject is an election, even a small error could lead a voter to misunderstand a candidate's position or a ballot measure's impact.
There is also the question of what data the chatbots were trained on and how current that information is. Election details change quickly, and a system relying on outdated sources may miss a dropped-out candidate, a shifted policy position, or a newly introduced local measure.
The rise of AI-assisted voting research arrives alongside broader debates about the role of technology in democratic life, including concerns about misinformation, deepfakes, and the influence of social media. Adding chatbot conversations to that mix introduces a new layer for voters, campaigns, and election officials to consider.
For now, the phenomenon appears to be driven by individuals rather than institutions. Voters like Johnson are experimenting on their own, treating the chatbot as one more input among many. Whether that experiment improves civic engagement or introduces new confusion may depend on how carefully users verify what the machines tell them.
What is clear from the NPR report is that the 2026 midterms are unfolding in an era when an AI conversation on the porch is, for some Americans, becoming part of how they decide.
(0)