Choosing the Next President: How AI Could Tip the Balance in Battleground States

  • National Newswatch

California Gov. Gavin Newsom speaks during an interview in the spin room before a presidential debate between President Joe Biden and Republican presidential candidate former President Donald Trump in Atlanta, Thursday, June 27, 2024. (AP Photo/John Bazemore)

Publisher’s Note: This column is the latest in a series by Don Lenihan exploring the issues around the use of AI, including the social, economic and governance implications. To see earlier instalments in the series, click here.

Two hundred and forty million voters are eligible to cast ballots in the 2024 US Presidential Election, but the outcome will hinge on a few hundred thousand votes in a handful of swing states. AI could be the decisive factor. Such is democracy in our time.

Both candidates are experienced AI users, for better or worse. In 2016, the Trump campaign famously hired Cambridge Analytica to “microtarget” voters on social media with data illicitly scraped from 50 million Facebook users. By contrast, the 2020 Biden campaign was a model innovator and regards AI as a critical factor in its win.

This election is different, not just because of the high political stakes. Generative AI (Gen AI) is reinventing political marketing. Campaigns can create highly personalized messages using text, images, and videos, targeting undecided voters on a massive scale with pinpoint accuracy. These messages can be adapted in real time, even as voters access them. Experts call this hyper-personalization.

Given the critical role of swing states in the Electoral College, this new capacity for “digital door-knocking” could determine the next President of the United States – and the future of democracy.

Electoral College and the Importance of Swing States

Despite the regular media reports of Trump or Biden closing the gap as they gain or lose a couple of points in national polls, this horse race means little. Americans vote for state electors who, in turn, choose the president through the Electoral College—the same process Trump tried to hijack during the January 6th riots.

Most states have been predictably "red" or "blue" for decades, and a small shift in the popular vote is unlikely to change their Electoral College outcome. The real contest is in the “battleground states,” which currently include Arizona, Georgia, Michigan, Nevada, Pennsylvania, and Wisconsin.

These states are pivotal – and incredibly close. In 2020, Biden won all six with a combined margin of less than 142,000 votes (Pennsylvania counted for 80,000). With 44 states effectively on the sidelines, shifting as few as 5,000 votes in some of these key battlegrounds could determine the result. Hyper-personalization—repeatedly reaching out to these voters throughout the campaign—could be a game-changer. We can see why by taking a closer look.

The Role of AI and the New Focus on Hyper-Personalization

Traditional microtargeting uses AI to (1) segment audiences, (2) predict behavior, and (3) tailor messages to specific groups or individuals. For example, Cambridge Analytica (CA) used Facebook data to build detailed psychographic profiles of millions of voters based on their recorded actions, beliefs, and preferences. CA then microtargeted individuals with messages designed to manipulate emotions and behaviors, often provoking fear or anger about an issue. While this approach can be effective, it is also unreliable, as a voter’s historical record may not accurately reflect their current opinions.

Gen AI, however, supports a “contextual” approach by creating new content instantly—text, images, and videos—and analyzing vast amounts of data in real time. Instead of relying solely on historical data, it focuses on what a person is currently doing to guide its response. For example, if someone is reading an article about labor shortages, the AI can recognize this and respond by presenting an ad about a party’s support for higher immigration levels. This results in highly nuanced, contextually relevant messages tailored to a voter's current interests and behaviors.

Hyper-personalization takes the best of both worlds. It uses traditional and Gen AI to leverage extensive historical data while also following and responding to real-time activities. For instance, if someone is browsing health-related material, the AI can review their historical data, revealing they are an undecided voter interested in healthcare reform. The AI then sends them a personalized message about a candidate's healthcare policies, effectively increasing the likelihood of voter engagement.

The takeaway here is that Gen AI is transforming marketing approaches. As one expert observes, “This shift [to hyper-personalization] is not just a trend but a significant movement.” Political marketing is on the same path, with major firms already lining up to advise political campaigns.

Deepfakes, Disinformation, and Opportunities with AI

As the 2024 election approaches, both parties are set to use AI to target undecided voters in battleground states. Tensions are high, and they will rise even higher. Deepfakes could become a significant issue. Imagine fabricated images, videos, or audio depicting Trump using racial slurs or Biden stumbling over his words. While such blatant attempts might be caught by national media, hyper-targeting offers a subtler approach.

By targeting small groups or individuals, deepfakes and disinformation can be disseminated with minimal risk of wide detection. These messages can be finely tuned to specific issues, particularly in battleground states where undecided voters are crucial.

On the other hand, hyper-personalization also presents opportunities for building more interactive and ongoing relationships with voters. AI chatbots can provide personalized interactions, answering questions and discussing policy positions based on a voter’s interests and concerns. This continuous engagement can build stronger, more trusting relationships with voters.

Moreover, AI can help campaigns track voter sentiment in real time, allowing for dynamic strategy adjustments. This responsiveness enhances voter satisfaction and loyalty, as campaigns can quickly address concerns and highlight relevant issues. By leveraging AI to create personalized, ongoing interactions, campaigns can build deeper connections with voters, ultimately strengthening democratic participation.

And Next?

The fact that a few hundred thousand votes in a handful of states will decide the presidency is a troubling commentary on modern democracy. While political parties are tight-lipped about their use of hyper-personalization in battleground campaigns, it's clear they will leverage these tools to the fullest in the decisive battle for the White House.

How will this play out? Will the AI strategies cancel each other out, or will one party gain a decisive edge? Could illicit practices like deepfakes or disinformation tip the scales, or will other issues, such as Joe Biden’s age or Donald Trump’s legal troubles, dominate the public narrative?

The outcome remains uncertain, and the stakes are incredibly high—not just for the candidates and voters, but for democracy itself.

Don Lenihan PhD is an expert in public engagement with a long-standing focus on how digital technologies are transforming societies, governments, and governance. This column appears weekly.