Exploring the artificial intelligence “Trust paradox”: Evidence from a survey experiment in the United States