Machine logic can be too precise to seem fully human. Chatbots that learn logic from Aristotle instead might behave more like people.