By David Kanevsky (President and Founder of 3D Strategic Research)
Note: With the ever-expanding debate on the use of A.I. within our industry, we have decided to bring the discussion to our newsletter and highlight opinions from both perspectives. Our first piece in this series came from Caryn Clark, a voiceover artist we have partnered with many times over the last few cycles, where she cautioned against A.I. Today’s piece on the counterargument is from David Kanevsky, President and Founder of 3D Strategic Research, highlighting his argument on the use of A.I. responsibly.
The practical question political consultants should have is not whether or not to use A.I. in campaigns, but how to use it well and responsibly. Campaigns have been working with machine learning algorithms for years under labels like predictive modeling and programmatic advertising. Generative A.I. tools are just the next step in that evolution.
Most political consultants are already using A.I. According to a bipartisan study I helped conduct on behalf of the American Association of Political Consultants, 86% of campaign operatives used A.I. during the 2024 cycle, with a majority using it at least a few times a week. Most consultants use A.I. to draft internal products or for creative ideation. While, like other technologies before it, A.I. will bring disruption, its most effective uses will be to augment people by making them more effective and efficient at scale.
Campaigns should consider A.I. as “Assisted Intelligence” rather than “Artificial Intelligence.” That means treating A.I. like a smart intern that needs training and guidance, while ensuring quality control over its output. For example, I will use A.I. to speed up coding of open-ended questions in surveys, then review and correct it since A.I. models can misclassify obvious answers when analyzing survey open-ends.
Poor use of A.I. in creative advertising is where misuse is most apparent. If you ask a general-purpose model like ChatGPT or Grok to write an ad or fundraising appeal, it will simply echo what is already out there. That yields cookie-cutter scripts and emails/texts, which is already a problem in a cluttered and fragmented media landscape when it comes to persuading voters and earning donors’ trust. Critics have called A.I. a “plagiarism machine,” and campaigns should take care since multiple candidates from both parties have had to deal with scandals for lifting copy on their websites and other material.
The solution is to spend time training an A.I. in a candidate’s tone and style by giving it examples of a candidate’s own speeches, writing, and content. That way, the A.I. will echo a candidate’s own words, rather than recycling the internet’s generic language. This can help speed the approval process for future creatives, while also scaling content for all the different mediums and formats a campaign needs to communicate in for the digital age.
Caryn Clark’s article about voter skepticism toward A.I. is real, but it’s just as important to look at how voters behave in addition to what they say about A.I. Randomized control trials from Grow Progress and Trilogy Interactive show cases where an A.I. voiceover matches or beats a human voiceover in persuading voters, as well as instances where it underperforms. While Grow Progress’ conclusion states that “AI voices can be a viable option for creating persuasive ads,” the data shows A.I.’s effectiveness can depend on the issue being discussed, the emotional content of the ad, the gender of the voiceover, and the audience being targeted. Just like deciding which human voiceover artist to use for an ad is often context dependent, so is the debate over whether to use an A.I. voiceover at all. That means consultants should not rely just on what they have done in the past or on their assumptions, but should test and validate what is most effective.
Additionally, voter attitudes toward A.I. will likely shift as adoption grows. In a national survey I conducted in November 2024, voters were more anxious and concerned (43%) than hopeful and optimistic (21%) about A.I. But among the 18% who use A.I. at least once a week, those numbers are reversed, with those more optimistic (45%) about A.I. outnumbering those who are concerned (23%). That pattern suggests skepticism will moderate as voters use A.I. tools in their daily lives, just as online shopping moved from a novelty to normal.
Political campaigns are an arms race. Every cycle, one side pioneers a tool, and the other side adapts and builds on it. The same will be true with A.I. The consultants who learn to use it well, with rigorous testing and human oversight, will get better insights, make smarter targeting decisions, and deliver more tailored creatives, while doing it faster. The ones who refuse will watch their opponents do it for them.
