Results from the AI x Democracy Research Sprint — LessWrong