People Are Asking ChatGPT to Help Them Decide How to Vote in the Midterms

My take: NPR reported something I find revealing about how we're using AI today: voters in the United States are asking ChatGPT, Gemini and Claude to help them understand their ballot, compare candidates and, in some cases, decide who to vote for. A software company vice president in Nashville said he cross-checks information across several models to catch hallucinations, and a voter in Pennsylvania used Claude to build a nonpartisan site that helps others research candidates using official sources. A Johns Hopkins researcher flagged something key: chatbots tend to reinforce what you already think instead of challenging you to see it differently.

This isn't a problem unique to elections, it's the same challenge AI always presents: it's a tool that makes you more efficient at finding and organizing information, but it doesn't think for you and it isn't neutral by default. If you ask it in a way that already hints at your opinion, it will likely hand that opinion right back to you, confirmed. The same applies when you ask it to help pick a vendor, evaluate a job offer or research an investment.

What does strike me as a good example of responsible use is cross-checking sources, like the software VP did, or building something that helps others reach their own conclusions with verifiable data, like the site built with Claude. If you use AI to make an important decision, whether it's your vote, your business or your career, are you using it to think better, or to get it to confirm what you already wanted to hear?