AI Was Everywhere in the 2024 Election, But Deepfakes Were Only Part of the Picture

It was the biggest election year in human history: the year 2024 is the year of the “big cycle” when 3.7 billion eligible voters in 72 countries had the opportunity to go to the polls. These are also the first AI elections, with many fearing that deepfakes and false information generated by artificial intelligence could flood democratic processes. As the year 2024 draws to a close, it is instructive to look at how democracy is doing.

In a Pew survey of Americans from earlier this fall, nearly eight times as many respondents expected AI to be used for the worst purposes in the 2024 election as those who thought it would be used mostly for good. There are real concerns and risks in using AI in electoral politics, but it certainly hasn’t been all bad.

The dreaded “true death” has yet to happen – at least, not because of AI. And candidates are enthusiastically embracing AI in many areas where it can build, if used responsibly. But because all this happens within the campaign, and mostly in secret, the public often does not see all the details.

Communicating with voters

Another impressive and profitable use of AI is language translation, and campaigns have begun to use it more widely. Local governments in Japan and California and prominent politicians, including Indian Prime Minister Narendra Modi and New York City Mayor Eric Adams, are using AI to translate meetings and speeches for various audiences.

Even if politicians themselves don’t talk about AI, their constituents may use it to listen to them. Google released free translation services in 110 other languages ​​this summer, available to millions of people in real time through their smartphones.

Some candidates have used the conversational power of AI to connect with voters. American politicians Asa Hutchinson, Dean Phillips and Francis Suarez posted their interviews on their first presidential campaigns. Candidate Jason Palmer beat Joe Biden in the American Samoa primary, not least because of the use of emails, texts, audio and video generated by AI. Pakistan’s former prime minister, Imran Khan, used AI of his voice to deliver speeches in prison.

Perhaps the most successful use of this technology was in Japan, where the secretive and independent Tokyo gubernatorial candidate, Takahiro Anno, used an AI image to answer 8,600 voters’ questions and was able to come in fifth place among a highly competitive field of 56 voters.

Nuts and bolts

AIs have been used in political fundraising. Companies like Quiller and Tech for Campaigns market AIs to help write fundraising emails. Some AI systems help candidates target specific donors with personalized messages. It’s very difficult to measure the impact of these types of tools, and political consultants are confused about what actually works, but there is clearly interest in continuing to use these technologies in campaign fundraising.

Voting has been highly statistical for decades, and pollsters are constantly incorporating new technology into their systems. Techniques range from using AI to distill voters’ sentiments from social media – something known as “social listening” – to creating artificial voters who can answer tens of thousands of questions. Whether these AI applications will result in more accurate polling and campaign strategy insights remains to be seen, but there is promising research fueled by the ever-growing challenge of reaching real people through polls.

On the political planning side, AI assistants are used for a variety of purposes such as helping write political messages and strategies, producing advertisements, writing speeches and helping coordinate campaigning and get-out-the-vote efforts. In Argentina in 2023, both presidential candidates are using AI to create posters, videos and other campaign materials.

In 2024, almost the same power is used in various elections around the world. In the US, for example, a politician from Georgia used AI to generate blog posts, campaign photos and podcasts. Even mainstream productivity software suites like those from Adobe, Microsoft and Google now include AI features that are inevitable – and perhaps most useful for campaigns. Some AI programs help mentor candidates for senior positions.

Fakes and counterfakes

And there was false information and propaganda created by AI, although it was not as catastrophic as feared. In the days before Slovakia’s 2023 elections, fake news about electoral fraud is spreading. This sort of thing happened more often in 2024, but it’s unclear if anything had any real impact.

In the US presidential election, there was a lot of media coverage after Joe Biden’s fake voicemail told New Hampshire voters not to vote in the Democratic primary, but it didn’t seem to make much of a difference in that vote. Similarly, AI-generated images from hurricane sites did not appear to have much effect, nor did AI-generated celebrity endorsement sequences or viral deepfake images and videos that distorted the actions of candidates and appeared to be designed to capture their political weaknesses.

AI has also played a role in securing the information ecosystem. OpenAI used its AI models to disrupt Iran’s foreign influence efforts aimed at sowing division before the US presidential election. While anyone can use AI tools today to generate convincing fake audio, images and text, and that ability will always exist, tech platforms use AI to automatically rate content as hate speech and extremism. This is a good use, which makes content moderation more efficient and saves people from reviewing the worst cases, but there is room for it to be more effective, more transparent and more balanced.

AI models are more likely to grow and adapt to more languages ​​and countries than human presidential organizations. But the implementation so far on platforms like Meta shows that a lot of work needs to be done to make these systems efficient and effective.

One thing that didn’t matter much in 2024 was the ban on corporate AI developers from using their tools in politics. Despite market leader OpenAI’s insistence on preventing political use and its use of AI to automatically reject a quarter-million requests to produce images of political candidates, the company’s enforcement is ineffective and actual use is widespread.

The genie is free

All these trends – good and bad – are likely to continue. As AI becomes more powerful and capable, it is likely to permeate all aspects of politics. This will happen whether the AI’s performance is superhuman or subhuman, whether it makes mistakes or not, and whether the balance of its use is good or bad. All it takes is one team, one campaign, one external team, or even one person to see the benefit in automation.

Bruce Schneier, Senior Lecturer in Public Policy, Harvard Kennedy School and Nathan Sanders, Affiliate, Berkman Klein Center for Internet and Organization, Harvard University

This article is republished from The Conversation under a Creative Commons license. Read the first article.