How election experts are thinking about AI and its impact on the 2024 elections

Nearly half the world’s population are expected to vote in 2024 as more than 60 countries around the globe hold elections. Looming over those votes is the potential use of artificial intelligence to disrupt campaigns and elections.

Scott Warren, a visiting fellow at the SNF Agora Institute at Johns Hopkins University and a visiting fellow at the German Marshall Fund (GMF), recently attended a convening of global election officials held in Brussels. There, he heard about their efforts to prepare for an AI-fueled election season.

Image caption: Scott Warren

Last year, Warren and the SNF Agora Institute partnered with the R Street Institute to develop a set of conservative principles for building trust in elections, and he notes the use of artificial intelligence has come up in those conversations as well.

What do you see as the threats and opportunities AI presents during this election season?

The big issue folks are talking about, obviously, is misinformation or disinformation. There are two big things to think about.

One, some of the bigger threats in terms of misinformation or disinformation are not necessarily the ones everybody is talking about—such as deep fakes of Trump or fake news articles. Some of it, and Arizona Secretary State Adrian Fontes demonstrated this at the Brussels forum, is lower-profile election officials being used for deep fakes.

The public is getting a bit more used to the idea of a deep fake with Biden, Trump, or a national official. But with people who aren’t as well-known but profess to have expertise in elections—that is a specific area ripe for disinformation.

AI offers a number of opportunities and solutions for providing more information to people about how to participate in where to vote. We can also use it to clean up voter rolls. But I think election officials are still very much at the early stages of what can be done.

How does the threat of AI’s impact on elections compare to the threat posed by social media?

AI will use social media— it’s just more sophisticated and can happen at a broader scale. When you’re using social media, you had to create online trolls, and they weren’t necessarily sophisticated.

With AI, there’s going to be a lot more opportunity to use deep fakes and create false personas. Four years ago, there wasn’t a path to do that. In 2015, there was an article on Facebook claiming the Pope endorsed Trump that was widely shared. With AI, you can now make a video of the Pope endorsing Trump. It’s a similar type of information, but a more advanced way of getting it across.

Is there anything about this topic that gives you hope?

People are talking about this threat much earlier than they were with social media. There’s bipartisan support for trying to figure this out early.

And I think there are some efficiencies with elections that AI will be able to solve for. If it’s used well and targeted in the right way, it can provide a lot more information to people about where and when to vote.

We’re getting in front of this very early, and there’s opportunity for people to harness it in positive ways.