The Golden Opportunity for American AI — LessWrong