The process of replacing social relations between humans with interactions with AI has highlighted the problem of the correctness of our understanding of what social interaction is and what role the sociocultural and biological components of human nature in it. Evolutionary neurosociology —a field that combines neurobiological and social variables to examine sociological issues from an evolutionary perspective— is best suited to clarify this question. From this perspective, the human-AI interaction represents an evolutionarily formed relationship among social contexts (defined by AI’s communicative role and the peculiarities of its social interpretation), mental states and processes (including social cognition skills and basic emotions), and the brain’s neurobiological mechanisms and patterns. This study suggests the probability of an evolutionary mismatch in human-AI interactions, arising when humans use evolved adaptive traits for purposes other than those for which they were originally selected. The mechanism of evolutionary mismatch likely relates to the functioning of the social brain, which creates a kind of illusion of social interaction. Furthermore, the currently promoted human-centered AI approach encourages forming authority ranking relations with AI, associated with the basic emotion of anger and fear. These relationships, in turn, are linked to brain activity that potentially contribute to reduced prosocial attitudes beyond human-AI interactions.