The views expressed by contributors are their own and not the view of The Hill

The first AI-enhanced presidential election

Visitors stop by the booth of Nvidia at the Apsara Conference, an annual cloud service technology forum hosted by Alibaba Group, in Hangzhou in eastern China's Zhejiang province Tuesday, Oct. 19, 2021. The Chinese government on Thursday, Sept. 1, 2022, called on Washington to repeal its technology export curbs after chip designer Nvidia Corp. said a new product might be delayed and some work might be moved out of China. (Chinatopix Via AP)

The contours of the 2024 presidential election were relatively clear — until AI showed up with its ability to fine tune individual messaging more than ever before.

President Biden will likely be the Democratic nominee, with a great list of accomplishments but lukewarm enthusiasm from independent voters. Former President Trump, with wall-to-wall coverage of multiple legal challenges over the next year, will freeze the Republican primary process. Ron DeSantis will mount a credible challenge, only to become the latest Florida governor to raise a boatload of money and be sunk by the Trump maelstrom.

Two relatively unpopular candidates – one perceived as too boring and the other as too crazy – will stumble into a general election that Biden will win again. 

But there are two inter-related wild cards that can change this scenario — one obvious, the other hidden.

The obvious wild card is a third-party effort by the savvy No Labels gang. They claim to be “working to build the launching pad of an independent Unity presidential ticket,” based on the belief that both parties have been captured by their extreme wings. But analyses of the effects of the No Labels effort make a convincing argument that it would benefit Trump.


A tremendous amount of money will be spent to promote, and undermine, any third-party campaign, which is likely to result in nothing more than raising the profile of those involved on either side of the effort. No legitimate candidate is going to purposefully step on the landmine being constructed by people outside the Biden or Trump worlds who just want to be perceived as inside, or at least relevant.

The hidden wild card, a repeat of the 2016 direct interference by Russia and non-state actors, could be enhanced by a third-party campaign and new technology. The 2016 manipulation was ham-handed, amateurish and a poor example of micro-targeting messages. But it worked because hardly anyone was expecting it, looking for it or reporting on it. In 2020 the same efforts had much less impact, first, because everyone was expecting it, looking for it and reporting on it, and second, because the general tide was so against Trump that micro-targeting remained just that: micro, and overwhelmed by the macro currents of the election.

But in 2024 there is a new factor. The deployment of large language models in federated machine learning known as artificial intelligence (AI) not only will enable an easier and more specific use of micro-targeting — but will also encourage the development of narratives and visual imagery so easily, and with so little cost, that every person could receive advertising designed just for them, for their issue, for that hour of that day.

AI micro-targeting will be based on the sophisticated analysis of mountains of data, delivered quickly and inexpensively, and altered repeatedly to account for hourly changes in individual search patterns, purchases and other information — all things that AI does better, faster and cheaper than humans. With this information, AI prompters can create precise, individualized 30-second ads specifically targeted to a voter. No one has to be plugged into the matrix when you are plugged into social media and spoon fed, if not force fed, the exact flavor you are craving.

Consider this: In just one week in September, 2022, three text-to-3D video image generators came to market. Google announced DreamFusion, Meta/Facebook announced Make-a-Video and Nvidia announced GET3D, providing just about anyone with the ability to create realistic images and videos.

In an election with two candidates who are perceived as flawed, where a greater percentage of likely voters than ever before claim to be independent and where there may be a well-funded third-party candidate who can change trendlines, AI-enhanced micro-targeting that can be used to create inexpensive and nearly instant video advertising targeted to one person will become significantly more important.

That does not mean the election will be run by faux sentient machines, let alone by bad actors paying for and feeding those learning machines. But it does change a single national contest into millions of small, individual elections. It may be that the “Trump is crazy” tide or the “Biden is too boring” tide are so great that none of those specific individual pushes and pulls will matter. But they are going to happen one way or the other, with consequences that are unknown, mostly unseen and, so far, barely reported.

Joe Andrew is the former chair of the Democratic National Committee, Dentons, the world’s largest global law firm, and Nextlaw. He has been an early-stage technology investor for more than 30 years.