Anthropomorphisation of personal artificial intelligence as a factor in cultural and anthropological risk among the student population

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Introduction. The active integration of personalised artificial intelligence (AI) systems into everyday and educational practices is transforming human-technology interaction, endowing it with qualities akin to interpersonal communication and stimulating users’ tendency towards anthropomorphism the attribution of human characteristics to AI. This issue is particularly pertinent among students, for whom digital agents are becoming not only learning tools but also potential sources of emotional dependency and educational risk. The question of which individual psychological traits (such as attachment style and aspects of self-concept) predispose individuals to profound anthropomorphism of AI and increase their vulnerability to developing dependent interaction patterns remains insufficiently explored. Aim. The present paper aims to identify the relationship between attachment style, self-concept characteristics specifically, the discrepancy between the actual self and the ideal self and the tendency to anthropomorphise personal AI among the student population. This will facilitate the identification of user profiles with varying risks of developing emotionally dependent interaction patterns and support the development of preventative strategies within the educational environment. Methodology and research methods. The study involved 629 students aged 18 to 24 years who were active users of AI services. The following instruments were employed: T. Leary’s Interpersonal Relationship Inventory (IRI) to assess the images of the “actual self”, “ideal self”, and “my virtual assistant (AI)”; and an adapted version of the Experiences in Close Relationships-Revised (ECR-R) questionnaire. Statistical analyses included cluster analysis (K-means method), analysis of variance (ANOVA), and the Mann–Whitney and Kruskal–Wallis tests. Results. Three profiles of AI perception were identified: instrumental-mechanistic (39.3%), selectively humanising (44.8%), and fully anthropomorphic (15.9%). Significant differences were observed between clusters in attachment styles ( p < 0.001) and the degree of intrapersonal conflict (actual self/ideal self discrepancy). A risk group, comprising 16% of the sample, was identified, characterised by a triad of anxious-ambivalent attachment, pronounced intrapersonal conflict, and a tendency towards full anthropomorphisation of AI. Scientific novelty. For the first time, the relationship between the anthropomorphisation of personal AI and deep-seated personality structures such as attachment style and self-concept has been empirically demonstrated within the context of educational risks. User profiles, differentiated by varying degrees of vulnerability to emotional dependency, have been identified. Practical significance. The research findings can be used to develop targeted preventative programmes and screening tools aimed at ensuring the psychological safety of the educational environment in the context of digitalisation.