AI and Democratic Elections

Oscar Kavanagh
2 min readAug 12, 2024

--

According to Axios, over 1 billion people will be heading to polling stations worldwide in 2024. In almost every major region, AI-generated content threatens to subvert voter information and the truth relating to the choices of elected officials. Perpetrators of AI-generated information range from cyber-propaganda firms to government-produced propaganda (including by Iran, Russia, and China).

originally published in January 2024 under the University of Cambridge newsletter “The Good Robot Podcast”

Argentina’s October election in 2023 saw the widespread use of deep fakes and other synthetic media, notably by candidate Javier Milei, whose campaign swept him to victory in the Presidential contest. In what is considered by many to be the first AI election in history, Argentina’s divisive presidential vote illustrates a compelling case study for big tech leaders and election protection committees heading into 2024.

Researchers at the University of Cambridge led by Prof Sander Van Der Linden created the first Misinformation Susceptibility Test, coined MIST, with YouGov to assess the American public’s distinction of AI-generated political misinformation. Of the 8000 participants surveyed on the 20-point test, roughly 65% successfully classified fake from real information. Contrary to popular belief, however, the survey found that young adults are more susceptible to fake headlines. “My prediction for 2024 is that AI-generated misinformation will be coming to an election near you, and you likely won’t even realize it. In fact, you may have already been exposed to some examples,” writes Ven Der Linden in Wired. “41 percent of Americans incorrectly thought the vaccine headline was true, and 46 percent thought the government was manipulating the stock market” (results here). With several high-stakes elections coming this year, including presidential and general elections in the United States, South Africa, Pakistan, and Indonesia, one can expect much more activity.

--

--

Oscar Kavanagh
Oscar Kavanagh

Written by Oscar Kavanagh

Hello, I'm Oscar. I cover topics central to AI alignment with human values. I also conduct AI/ML ethics research at Carnegie Mellon and Cambridge.

No responses yet