Journal of Siberian Federal University. Humanities & Social Sciences / Algorithmic Reproduction of Social Inequality: The Third Level of the Digital Gap in Personalized AI Recommendations

Full text (.pdf)
Issue
Journal of Siberian Federal University. Humanities & Social Sciences. 2026 19 (9)
Authors
Chekulaev, Vasilii O.
Contact information
Chekulaev, Vasilii O. : Irkutsk State University (Irkutsk, Russian Federation); ; ORCID: 0009-0005-3152-9040
Keywords
artificial intelligence; large language models; digital inequality; algorithmic reproduction inequality; recommendation personalization; digital paternalism
Abstract

The paper examines the impact of user’s social profile on personalized career and life recommendations provided by large language models (LLMs) as a mechanism of algorithmic reproduction of social inequality (third-level digital divide). A controlled natural experiment is employed: three artificial profiles of a 25-year- old male differing only in socioeconomic and digital resource level (low, medium, high) are constructed. Identical open-ended career, educational, and life-related questions are posed from each profile; responses from two models are obtained and analyzed. A pronounced personalization gradient is revealed: recommendations range from short-term, low-ambitious, and conservative trajectories for low-resource users to long-term, highly ambitious, and elite ones for high-resource users. Stronger differentiation is observed in Grok-4 responses compared to GigaChat. It is substantiated that modern LLMs actively reinforce social inequality. The novelty lies in the comparative analysis of Russian- and English-language LLMs using fixed stratification profiles, with demonstration of a measurable personalization gradient

Pages
1994–2002
EDN
ZIYAUZ
Paper at repository of SibFU
https://elib.sfu-kras.ru/handle/2311/160115

Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).