The author examined four groups of constraints on the applicability of artificial intelligence (AI). The first group includes material resources for storing large data sets, which impose resource limitations on AI. The second group includes the computational capabilities of software code, where the fundamental constraint of AI is the incomputability of the human brain. The third group includes a set of space-time conditions governing computational procedures, which unfolds into six types of constraints: AI as a time-closed system with a degraded future, AI as a time-closed system with both a degraded future and past, the heterogeneity of G. B. Minkowski’s space-time, the biological space-time of V. I. Vernadsky, Kairos, and real time. The fourth group includes the civilizational conditions of social development, where the limit of AI is defined by Vernadsky’s Noosphere. When this scheme of AI constraints is applied to human consciousness, it becomes possible to more precisely delineate the “boundary” between natural intelligence and human thought. Thought is defined as the aspect of consciousness that transcends mere intelligence and directs the development of humans and AI. A natural framework for the interaction between humans and AI emerges from this definition.