The problem of rationality at the border of expressiveness and possibilities of human thinking is discussed when comparing natural and artifi cial intelligence (AI). A distinction is made between the concepts of intelligence and thinking. At the same time, intelligence is defi ned as rationality, always working within the limits of expressibility, and thinking includes inexpressibility in its structure. It is possible to better differentiate intelligence and thinking through an appeal to AI by formulating groups of restrictions for it, which together with this are the limits of expressiveness. Four groups of AI limitations are considered: 1) the material support of big data storage, where we come to the resource limitation of AI; 2) the computational capabilities of the program code, where we come to the fundamental incalculability of the human brain; 3) a group of spatio-temporal conditions of the computational procedure, which is decompressed by 6 types of restrictions of this type. 3.1) AI as a time-closed system with a degraded future. 3.2) AI as a time-closed system with a degraded not only future, but also past. 3.3) Heterogeneity of G. B. Minkovsky’s space-time. 3.4) Biological space-time of V. I. Vernadsky. 3.5) Kairos. 3.6) Real time. 4) Civilizational conditions for the development of society, where the limit of AI is determined by the Noosphere according to V. I. Vernadsky. When transferring this scheme of limitations for artifi cial intelligence to human consciousness, it is possible to more accurately defi ne the ‘boundary’ between natural intelligence and thinking. Thinking is defi ned as a part of consciousness that is outside the intellect and determines the development of both the person himself and artifi cial intelligence. From this defi nition, a natural scheme of human interaction with artifi cial intelligence is derived.
Ilya Nikolaevich Volnov (Thu,) studied this question.