Artificial intelligence (AI) is penetrating our daily life and has become an important factor for research and economic growth. But with every new success story, false expectations are awakened again and again, and this follows a long tradition. Today, AI is mostly understood as machine learning, especially deep learning, whose recent successes are responsible for the current popularity of this term. However, due to some unique ability of an AI system, limited to a narrowly focused field of application, often unjustified generalizations about its actual "intelligence" are made. In this talk, we show that AI can not only be reduced to deep learning, but must also consider symbolic knowledge representation, as well as planning and optimization.