Oct 7, 2022 - Politics & Policy

AI becomes a political "super-weapon"

Illustration of a dollar sign made of binary code changing into a checkmark made of binary code.

Illustration: Brendan Lynch/Axios

Innovations in artificial intelligence are making it faster and cheaper for political campaigns to identify, turn out and extract money from voters.

The big picture: Consultants for both major parties are hoovering up voter data to hone advanced fundraising and persuasion tactics. These data tools are especially useful in down-ballot local races.

  • Consultants say they can dramatically improve campaigning by asking machines to synthesize huge amounts of data — from incomes to consumer purchasing habits — to predict how individuals will donate and vote.

What's happening: A host of consultancies are now marketing their use of AI and machine learning to boost political clients.

  • The Sterling Data Company, a Democratic firm, says it can more than double digital fundraising performance in the immediate term.
  • Numinar, a Republican startup, touts highly effective voter modeling and predictive capabilities.
  • Veteran Democratic fundraiser Anne Lewis' firm, MissionWired, recently unveiled a fundraising product called AdvantageAI.

What they're saying: "This is a super-weapon that Democrats have," boasted Martin Kurucz, Sterling's managing partner, who says his firm has worked with about 1,000 Democratic campaigns and political committees.

  • "It's probably one of the most overlooked reasons why the Democrats are winning the small-dollar fundraising wars against Republicans," he said.
  • Numinar says it's worked with nearly 300 political clients this cycle, and FEC records show payments from some key Senate and House campaigns this year.
  • "AI and machine learning can actually help these campaigns, 80% of which are down-ballot local races, make a pretty big difference in how their races perform," Will Long, Numinar's founder, told Axios in an interview.

How it works: Each company in the space uses massive amounts of political and consumer data to feed an ever-improving algorithm that can predict voter or donor behavior.

Sterling uses the tech to hone fundraising tactics tailored to each of its clients, the company says in its marketing materials:

  • "No data analyst in the world could look at over 500 variables, from household income to magazine subscriptions, and determine what factor played the biggest role in the success (or failure) of a fundraising campaign and rework the targeting to maintain or improve results — especially not in less than a minute."

AI-based software also makes it much easier for firms to identify new donors — people who are calculated to have the highest propensity to give to a campaign.

  • And Numinar says its ability to predict how a given voter will vote makes its services highly valuable to campaigns attempting to more efficiently allocate scarce resources.

Reality check: Some political pros are skeptical, describing it as more of a marketing gimmick than a true advance in political data technology.

  • "Some folks are using artificial intelligence as a game changer. And then there are a lot of people that are using artificial intelligence and machine learning in order to sell a product," said Mark Jablonowski, a Democratic digital strategist.
  • Asked about the trend, one veteran Republican consultant facetiously texted a GIF labeling it "magic." He described AI in the political world as "a catch-all BS word."

Yes, but: Numinar's Long acknowledged the term can be misused but said AI is an accurate descriptor of some of the highly advanced technology being deployed in political races across the country.

  • "We actually are using machine learning in a very core, fundamental way," he said, pointing to internal voter models that improve every time the firm feeds new data — provided by its political clients — into those models.

The bottom line: Whether you prefer to call it AI, machine learning or old-fashioned data-crunching, the technology underlying political campaigns is rapidly improving.

  • "Every election cycle, political data, analytics, and modeling grow by leaps and bounds," Jablonowski said. "The types of models created and the inputs that go into them increase tremendously."
Go deeper