AI Poised to Automate Election Persuasion, Sparking Preparedness Concerns
Concerns are growing that advancing AI technology could automate voter persuasion in elections and impact election outcomes, with experts pointing to a lack of preparedness for this political application of AI. MIT Technology Review reported on December 5 (local time) that AI is highly capable of political persuasion and has the potential to automate the process on a large scale, warning that preparations for the use of AI in elections are inadequate. AI can tailor arguments to individual voters, conduct tests to identify the most effective messages, and be used to subtly adjust political views. In the past, campaigns to exert influence online required a large workforce to create fake accounts and produce memes, but now AI technology can automate these tasks.
AI chatbots have been shown to have a much greater influence than traditional political advertising in changing voters' political views. Interaction with AI chatbots can change voters' mindsets and attitudes, which directly impacts election results. Examples of AI-driven election interference are already emerging. In January 2024, in New Hampshire, AI-generated fake calls mimicking the voice of U.S. President Joe Biden were sent en masse to Democratic supporters to discourage participation in the primary election. This incident is a prime example of the potential dangers of AI technology in elections. AI tools like OpenAI's Sora can be used to create messages from politicians or celebrities, fake news clips, etc., in minutes. These tools can be exploited to spread disinformation and confuse voters, posing a serious threat to the foundations of democracy. MIT Technology Review analyzed that it could cost less than $1 million to generate personalized, interactive messages for every registered voter in the United States.
In particular, AI-based personalized messaging campaigns targeting swing voters, who can determine election results, are even more threatening. MIT Technology Review analyzed that AI-based targeting of the approximately 80,000 swing voters who decided the 2016 U.S. presidential election could be achieved for less than $3,000, emphasizing that influence over a specific voter base can be maximized at a low cost. Election interference using AI can cloud voters' judgment and lead them to make choices based on misinformation. In addition, AI can be used to manipulate support for a particular candidate or spread negative opinions about an opposing candidate. AI chatbots can interact with voters to understand their political leanings and interests, and deliver personalized messages accordingly. These personalized messages can target voters' psychological vulnerabilities and appeal to their emotions to elicit support for a particular candidate or party. Creating and distributing fake news or videos using deepfake technology is also a form of election interference using AI. Deepfakes are so sophisticated that they are difficult to distinguish from reality, causing voters to believe false information and severely affecting election results. Experts emphasize the urgent need to prepare for AI-driven election interference. Active investment and effort by governments and related institutions are needed to keep pace with the rapid development of AI technology, and it is important to strengthen media literacy education for voters to increase their ability to distinguish between false information. Continuous research and analysis of the impact of AI technology on elections should be conducted to identify new methods and techniques of AI-driven election interference and develop countermeasures. In addition, legal regulations on AI-driven election interference should be strengthened, and penalties for violations should be increased to deter illegal election activities using AI.
