I admit to Hutchinson that if I were a politician, I would be afraid to apply BattlegroundAI. Generative AI tools are known to “hallucinate,” which is a polite way of saying that they sometimes invent things from scratch. (They bullshit(Using academic jargon.) I ask how he ensures the accuracy of the political content generated by BattlegroundAI.
“Nothing is automated,” he replies. Hutchinson notes that the BattlegroundAI copy is the starting point, and the campaign people are supposed to review and approve it before it ships. “You may not have a lot of time or a huge team, but you’re definitely reviewing it.”
There’s a growing movement, of course, to oppose the way AI companies train their products to do art, writing, and other inventive work without asking permission. I ask Hutchinson what she’d say to people who might object to the way tools like ChatGPT are trained. “These are incredibly legitimate concerns,” she says. “We need to talk to Congress. We need to talk to our elected officials.”
I ask if BattlegroundAI is considering offering language models that train exclusively on public domain or licensed data. “We’re always open to that,” he says. “We also need to make sure that people, especially those who are time-constrained, in resource-constrained environments, have the best tools available to them. We want to have consistent results for users and high-quality information—so the more models available, the better for everyone.”
And how would Hutchinson respond to opposition from progressives—who are generally pro-union—to automating advertising copywriting? “Valid concerns, of course,” she says. “The concerns that come with any new technology—we’re afraid of the computer, we’re afraid of the lightbulb.”
Hutchinson offers her take: She doesn’t see it as a replacement for human labor, but rather as a way to reduce drudgery. “I worked in advertising for a long time, and there are so many repetitive elements that, quite frankly, suck creativity out of it,” she says. “AI takes the boring stuff out of it.” She sees BattlegroundAI as a lifeline for overworked and underfunded teams.
Taylor Coots, a Kentucky political strategist who recently started using the service, describes it as “very sophisticated” and says it helps identify target voters and ways to tailor messages to reach them in ways that would otherwise be tough for diminutive campaigns. In battleground races in gerrymandered districts where progressive candidates are major underdogs, budgets are tight. “We don’t have millions of dollars,” he says. “Any opportunities to increase efficiency that we have, that’s what we’re looking for.”
Will voters care if the text in digital political ads they see is AI-generated? “I’m not sure there’s anything more unethical about AI generating content than about anonymous employees or interns generating content,” says Peter Loge, an assistant professor and program director at George Washington University who founded the Political Communications Ethics Project.
“If it were possible to mandate that all political texts written using AI be disclosed, then logically it would also be necessary to mandate that all political texts” — such as emails, advertisements and editorials “not written by the candidate” — be disclosed,” he adds.
Still, Loge has concerns about what AI is doing to public trust at a macro level, and how it might affect how people respond to political messages in the future. “One of the risks to AI is less about what the technology does and more about how people perceive what it does,” he says. “People have been falsifying images and inventing things for as long as politics has existed. The recent interest in generative AI has increased people’s already incredibly high levels of cynicism and distrust. If everything can be false, then maybe nothing can be true.”
In the meantime, Hutchinson is focused on his company’s short-term impact. “We really want to help people now,” he says. “We’re trying to move as quickly as we can.”