This paper deals with fundamental terms that are key for understanding the application of AI systems in public administration. First and foremost the problem of defining the very concept of artificial intelligence is presented, and then various subtypes of artificial intelligence are described. After that follows the part about algorithms, and the part about the importance of data for the sole purpose of application of AI systems in public administration. The central part of the paper contains prerequisites that are needed to be filled before the application of cognitive technologies in public administration, which is very important in order to avoid potential negative outcomes such as discrimination and bias. Before application it is of great importance to evaluate the state at which is the administrative organization in question, to define problems that are needed to be solved and goals that are wanted to achieve, and take in mind all of the benefits and the challenges of AI systems, so that the application would be successful and the negative outcomes would be reduced to a minimum. Furthermore, the paper briefly describes the efforts made so far in order to achieve regulation of the application of cognitive technologies in public administration, and attitudes of The Council of Europe, European Union, and Organization for economic cooperation and development regarding the application of artificial intelligence in the work of public administration.