AI is changing what political campaigns say, and who gets to hear it

For years, political campaigns have tried to learn everything they can about voters: What they drive. Where they shop. Whether they own a home, a boat or have a particular interest in equestrian sports.

The data helps campaigns make educated guesses about people’s politics, then decide which voters are worth pursuing and what might persuade them. It is a practice that predates social media, let alone ChatGPT.

But a new generation of artificial intelligence tools is shifting the art of political messaging away from broad-based appeals toward more carefully tailored micro-targeting.

Campaigns can now use AI to sift through research, study audiences and develop political materials in a fraction of the time those tasks once required. Generative technology can produce dozens of variations of an advertisement or appeal, while increasingly sophisticated algorithms help determine which messages reach which audiences.

The advances are arriving as Minnesota campaigns prepare for November’s election, raising questions about how extensively candidates are using the technology and what its growing presence means for voters. Much of the attention has gone to fabricated images and videos. Less visible is the work happening behind the scenes: the research, analysis and decisions about who should see what.

Nora Raube, who directed digital advertising for Sen. Amy Klobuchar’s 2024 reelection campaign, said the fixation on AI-generated content obscures the more consequential ways the technology is entering political work.

Campaigns can use specialized tools to research opponents, identify where potential audiences spend time online and analyze information that would otherwise require hours of manual work. The technology can help political operatives make decisions faster, often without revealing much about how those decisions were reached.

But how effective are the machine-assisted messages campaigns then produce? The research, so far, is mixed.