Fenwick McKelvey in MIT Press:
Could computer simulations someday predict elections worldwide with perfect accuracy? As far back as the early 1960s, Orrin Clotworthy certainly thought so. The CIA officer and computer hobbyist imagined that in the future, computers, then a fledgling technology, would not only predict electoral outcomes but explain them better than the smartest politicos. They would do this, he wrote, by “bypass[ing] the voter himself in this analytic process, looking beyond him for the reasons underlying his decisions.”
Today, Clotworthy is not alone in this dream. Simile, a $2 billion AI startup, is a “simulation company” promising to make the “most accurate behavioral models in the world by painting a living portrait of the unique people who shape it.” Like Clotworthy, Simile will use a “social simulacra” — in this case AI agents trained on interview data — to open the black box of human behavior and motivation and reveal the mystery of elections. They appear to be on the path to success; the firm’s website highlights a recent study that found their models “reproduced participant responses at 85% of human self-retest accuracy.”
More here.
Enjoying the content on 3QD? Help keep us going by donating now.