Why artificial intelligence must be explainable (XAI)?

Humans can hardly understand the algorithms used for machine learning and text-driven artificial intelligence applications: How and why decisions are made? Are the results fair,