Artificial intelligent systems and the problem of experience

The development, implementation, and further improvement of artificial intelligence (AI) systems tend to be related to the problem of experience. Unlike programs as closed algorithms, such systems interact with the environment and are able to change it in practice. Thus, contemporary discourse appears to ascribe «abilities», «learning skills», «decision-making», etc. to artificial agents. However, is it correct to extrapolate to artificial intelligent systems (AIS) the sense of phenomena characteristic of living beings? Is a machine truly able to gain experience, learn, and make decisions? The search for answers to such questions encourages investigation into the category of experience, its structure, and its gaining by living beings. Due to the ambiguity of this category itself, the use of the phenomenological approach seems to be the most productive. It allows us not only to clarify the essential features of experience but also to explore its multidimensional connections with practice, memory, imagination, will, setting and achieving goals. An analysis of specific examples also helps in evaluating analogues of these components for artificial agents and in systematizing problems that arise with further improvement of AIS. The presented results show that the category of experience, in the strict sense of the word, is not applicable to the currently functioning «weak/narrow» AI. However, the possibility of modeling this phenomenon seems to be open within the framework of future developments of «strong/general» AI. In conclusion, the reader may find factors that should be taken into account and implemented when creating AIS that would be able to gain experience and to carry out conscious practical activity.