Are Democrats Ready for Our First Truly AI Election?

"Every day that we are not adopting this stuff, the other side is moving faster. And I think ultimately that's going to translate into votes." — Jack Welty This is the first election cycle where what ChatGPT, Gemini, and Grok say about candidates could actually shape how voters decide. And as Jack Welty sees it, our side isn't ready. Jack served as Deputy Chief Analytics Officer on the Harris for President campaign and led paid digital analytics for Biden-Harris in 2020. He co-authored a paper on practical AI adoption for campaigns and is the co-creator of Caucus AI, a tool that monitors what AI chatbots are telling voters about candidates across more than 1,100 races. He joined me on this episode of Hello Merge Tag to talk about something I don't think our side is taking seriously enough. The conversation covered a lot of ground. But the through line was this: Democrats are falling behind Republicans on AI adoption, and the gap isn't closing. The American Association of Political Consultants found that 44% of Republican consultants use AI daily. Only 28% of Democrats do.  A few things that stuck with me: The biggest wins from AI aren't the sexy stuff. Jack's argument is that the most underrated use of AI on campaigns is automating the boring, high-volume work that eats up staff hours. On the Harris campaign, his team fed thousands of post-shift volunteer feedback surveys into a language model and got back structured reports by date, state, and issue type. What would have required an engineer and custom code 18 months ago, he says you could now do by dragging a folder into Claude. AI is a management challenge, not a technical one. This reframe is the most useful thing I took away from the conversation. You don't need to know how to code. You need to know how to delegate, review, and give feedback. The analogy Jack uses — treating AI like an eager but inexperienced junior staffer — is exactly right. You wouldn't hand an intern a blank page and say "write me a report, see you at five." You'd give them examples, context, and check in along the way. Same deal here. Which chatbot a voter uses quietly determines which media ecosystem they're in. Caucus AI tracks what ChatGPT, Gemini, and Grok tell voters about candidates. The responses are broadly factual — nobody's telling voters who to pull the lever for. But the sources each model cites are very different. Grok cites social media. ChatGPT leans heavily on its national media partnerships — Axios, AP, Washington Post. Gemini casts a wider net and is more likely to cite local news. Same question, three different information worlds. -Your .gov site matters more than you think. This is something I had definitely been sleeping on. Caucus AI found that official government sites get cited by these models at a much higher rate than campaign sites. If you're an incumbent and your official site is a graveyard of four-year-old press releases, you're leaving something real on the table. Same goes for Ballotpedia — candidates can fill out their own candidate surveys, and Ballotpedia is one of the top two or three cited sources across most of these models. Most campaigns aren't doing this. They should be. ChatGPT indexed a brand new Wikipedia page in 12 minutes. Caucus AI created a Wikipedia entry for a lesser-known candidate and within 12 minutes, ChatGPT was citing the brand new page. Two-thirds of all sources in its subsequent answers about those candidates came from that single page. There's no fact-checking layer. No delay. It's pretty clear your campaign needs a Wikipedia strategy! Find links, transcript and more hellomergetag.com.