The Hoover Institution at Stanford University conducted a study of the biases of Large Language Models of artificial intelligence and found that all models skewed left.