In a groundbreaking development, two 19-year-old college dropouts, Cam Fink and Ned Koh, have successfully predicted the outcome of a highly anticipated New York Democratic primary election using AI chatbots, without surveying a single human voter.
Their startup, Aaru, claims to have cracked the code on election polling by replacing traditional methods with thousands of AI agents. This innovative approach delivers faster and, they argue, more accurate results.
Aaru’s AI-driven poll predicted centrist George Latimer’s victory over incumbent Jamaal Bowman with near-perfect accuracy, forecasting a 58.7% to 41.3% split — closely matching the final tally.
Aaru’s methodology leverages census data to simulate voter behavior, gaining attention from high-profile entities, including Fortune 500 companies, political campaigns, and think tanks. One California campaign now relies primarily on Aaru’s polling for its strategic decisions.
The AI agents, programmed to mirror voter personalities, adapt their political preferences in real-time as they consume media. Each poll draws on 5,000 AI respondents, producing results in under a minute at a fraction of the cost of traditional polling.
The AI agents continuously browse the internet to mimic the media habits of the voters they represent. This process can lead to shifts in voting preferences, as seen when many agents initially shifted their support to Donald Trump following an assassination attempt, only to revert after further details emerged about the incident.
Despite skepticism about the accuracy of large language models, Aaru’s success suggests AI-based polling could become the new standard. The think tank Heartland Forward recently hired Aaru to conduct a survey on AI perceptions across nine U.S. states, further indicating the growing interest in this groundbreaking technology.
Cam Fink is confident in the future of AI polling. “No traditional poll will exist by the time the next general election occurs,” he stated.
