The concept of reasonableness in relation to artificial intelligence is a subject of scientific and philosophical debates, since there is no single universally accepted definition of the concept of “intelligence” itself. However, it is possible to identify key aspects that allow us to assess the potential capabilities of artificial systems to demonstrate the reasonableness of behavior. Obviously, the criterion of artificial intelligence reasonableness should include ethical and functional evaluation aspects. Artificial intelligence is not only about imitating the human way of cognition, but also about achieving rationality and optimizing the processes of decision-making, goal setting and achievement. Clear criteria for evaluating artificial intelligence systems are crucial to ensure that these technologies are useful, ethical and consistent with societal values. This implies a balance between innovation and regulation. The proposed concept is based on an original approach to assessing the artificial intelligence reasonableness through the prism of its ability to independently generate the purpose of activity and reproduce the most relevant to expectations result, as well as to ensure the reproducibility and replication of this result.