Mastering Trustful Artificial Intelligence

As a counter-thesis to the naive general narrative that artificial intelligence (AI) is a hyped super-technology which can solve all problems and even overtake human intelligence, this article discusses five essential problem areas associated with AI technology: modeling ability, how do we derive models of the real world from data and how do we create a model without prejudices and errors?; verifiability, how do we verify the AI algorithms?; explainability, how can we understand the decision-making process of AI systems?; ethics, how do we guarantee compliance with ethical principles and values?; and finally, responsibility, who is responsible for the decisions made by the AI system? We also discuss fundamental threat scenarios in the context of our information society, as well as the limits of AI technology compared with human intelligence. The article highlights that the development of AI technology, as well as related policies and regulations, must be organized to ensure its socially acceptable use and rule out any improper use. Furthermore, the article provides a philosophical discussion on the limits of AI and the diversity of life, and shows that we bear the ultimate responsibility for what machines do. This paper concludes by arguing that even the most sophisticated machine will probably never be able to match humans in terms of their multi-dimensionality of cognition, emotion, and physicality and in their sensual perception of the world.