AI Heralds New Era of Election Prediction with Virtual Voters

Artificial intelligence (AI) has emerged as a new variable in the landscape of nationwide local elections. Simulations that utilize AI to craft election strategies and predict outcomes, going beyond simple opinion polls, are spreading.

In the past, reliance was placed on statistics-based opinion polls. However, recent AI technology has broadened the scope of election analysis by comprehensively integrating and interpreting vast amounts of online information. Driven by advances in generative AI, research is emerging that simulates election results by assigning characteristics of virtual voters to large language models (LLMs).

Researchers at Cornell University in the United States have developed the 'FlockVote' framework, which creates a 'virtual voter society' composed of multiple LLM agents to simulate elections. This research team simulated voting outcomes in seven swing states for the 2024 US presidential election and reproduced patterns similar to the actual election results.

AI has become an essential component for election campaigns, with each political party utilizing AI across strategy development and voter communication for the June 3 local elections.