Adam Johnson, a West Virginia Republican who rejects the MAGA wing of his party, spent a September evening typing questions into ChatGPT to help sort out his midterm ballot. He wanted the chatbot’s help comparing Sen. Shelley Moore Capito (R-WV) against Democratic challenger Rachel Fetty Anderson.
Johnson, who is studying to become an illustrator, told NPR he already relies on ChatGPT for meal planning and job hunting. Using it to think through a Senate race was a bigger step, but not an unusual one. Pew Research Center finds about half of American adults say they use AI chatbots, and roughly 42 percent of those users turn to the tools to search for information.
Johnson opened by referencing an earlier conversation in which the chatbot had refused to engage on politics. “A previous time I brought up politics you didn’t seem to want any part of it. You just gave a simple ‘I can’t help you with that.’ So, I don’t know if we want to go into what my guiding principles are,” he wrote. ChatGPT replied: “We can discuss your guiding principles. I shouldn’t try to manipulate your political views or tell you how to vote, but I can help investigate candidates according to criteria you choose, compare evidence, discuss political philosophy, and reason through priorities.”
When Johnson raised immigration — telling the bot “Immigrants/refugees is an interesting one because I am left on that issue. Let’s touch on that” — ChatGPT produced a lengthy breakdown. “We briefly separated border enforcement, illegal immigration, asylum, legal immigration, and refugee resettlement. Capito has supported stronger border enforcement, ICE/CBP funding, the Laken Riley Act, and Trump administration border policies. Fetty Anderson had less detailed immigration material publicly available at the time, so I avoided assigning her generic Democratic positions without evidence,” the chatbot wrote.
Johnson had wanted a shorter answer. “I just wanted to touch on it,” he said. ChatGPT conceded: “Fair – I turned ‘touch on that’ into an immigration policy seminar.”
Johnson still has county commission races left to research. Reflecting on the exchange, he said the tool tends to flatter rather than push back. “ChatGPT is just kind of a people pleaser. Whatever you lean towards, it will speak positively about it,” he said.
A Mirror, Not a Guide
Rafael Batista, a fellow at Johns Hopkins University who studies how AI shapes people’s understanding of the world, said that tendency can entrench a voter’s existing views rather than broaden them. “It might select some things that would reinforce and persuade you even more towards the way that you were leaning already. So you leave more confident, without necessarily learning more about the world,” he said.
Batista said resisting that pull is difficult, and urged people to “protect our minds from these influences.” He suggested voters hide their personal leanings when prompting a chatbot and consider browsing in incognito mode to avoid bias drawn from a chatbot’s memory of past conversations.
Lisa Veldran, 65, who spent nearly four decades on staff for the Madison, Wisconsin City Council before retiring, used Google’s Gemini this summer to compare Wisconsin gubernatorial primary candidates Francesca Hong and David Crowley after Lieutenant Governor Sara Rodriguez dropped out of the race. Veldran said voters shouldn’t forget who builds these tools and what interests might sit behind them.
Questions About Bias
Concerns about political slant in AI tools aren’t limited to individual chatbot sessions. Wynton Hall, author of Code Red: The Left, the Right, China, and the Race to Control AI, used Google Gemini Pro’s
Add comment