How are you deciding who to vote for in the coming midterm elections? Many Americans are turning to ChatGPT.
In a new piece, NPR spoke to several voters who use the OpenAI chatbot to research candidates and get a rundown on their voting records. Some are asking the AI to create spreadsheets on different races or to verify or debunk viral claims. One even used AI to create an entire website for retrieving voting info.
Experts are aghast, noting that AI models tend to be sycophantic and aren’t as impartial as they seem.
“It might select some things that would reinforce and persuade you even more towards the way that you were leaning already,” Rafael Batista, a researcher at John Hopkins University, told NPR. “So you leave more confident, without necessarily learning more about the world.”
40-year-old Adam Johnson in West Virginia uses ChatGPT to organize his life, and he’s now also using it to prep for the midterms. Like many voters, he harbors a grab bag of beliefs. He votes Republican unless the candidate is MAGA, in which case he votes Democrat or third party. He also identifies as “left” on immigration and refugees.
When he asked CHatGPT why it avoided discussing politics before, it gave a mealy-mouthed explanation for why it was now venturing into such territory.
“We can discuss your guiding principles,” it began, per NPR. “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.”
Lakshmi Iyer, 50, said she used Claude to create a website where Pennsylvania voters can learn about and learn the specific stances of candidates. She caveated that she was aware that if AI “can infer your political leanings, they will push you towards what you are leaning to.” Johnson also acknowledged AI’s sycophancy, calling ChatGPT a “people pleaser.”
“Whatever you lean towards, it will speak positively about it,” he told the NPR.
Still, it’s one thing to acknowledge a risk and another to meaningfully internalize it. Much research has been published on how AI models encourage “cognitive offloading,” in which people more or less outsource their thinking to an AI, and it’s not hard to imagine something similar happening with civic engagement — rendering users, in a worst case scenario, as little more than “meat proxies” who vote how the AI tells them to.
It becomes a more alarming trend when you consider AI companies’ deep political entanglements. OpenAI is linked to PACs that have donated tens of millions of dollars to both Republican and Democrat candidates. Its president Greg Brockman donated $25 million to a Donald Trump Super PAC in 2025 and remains closely aligned with the administration. That’s not to mention the other ways that OpenAI and other AI companies are intertwined with the status quo, such as meeting with Trump to form a cartel-esque agreement to slow down and “self-regulate” AI development.
Of course, you may shop at a store whose owners heavily support a particular candidate or movement. But that store isn’t effectively telling you how to think and determining what information you see. Are we simply offloading our brains to determine our politics, or are we inadvertently being puppetted by AI models controlled by companies that act in their self-interest? Or maybe, with all their toadying, AIs are herding us towards the extreme ends of our own political biases we didn’t even know we had, exacerbating the already toxically polarized political environment.
More on AI: Sam Altman Says the World Is Going to Have to Suffer for the Glories of AI
The post Particularly Weak-Minded Americans Are Asking ChatGPT Which Candidate to Vote For, Which Sounds a Bit Like AI Is Just Voting Now Via Meat Proxies appeared first on Futurism.
