Today I discuss The Middle Principle, where humans should start and end all AI interactions. We should use direction at the beginning to ensure we give AI the best possible input, then use judgement at the end to check the output is accurate, correct and fits our needs.