The role of social media algorithms in fueling radicalization in Pakistan

Abstract

This research paper reviews key considerations at the intersection of AI, digital recommender architectures, and socio-political stability in the Global South, with a detailed focus on how social media algorithms can exacerbate radicalization in Pakistan. The digital sphere in Pakistan has registered dramatic growth in scope over the last ten years, and it has essentially become an algorithmically mediated, polarized arena. Commercial considerations focus on engagement levels, user retention, and monetization of emotional reactions, which drive the structural design of algorithms to promote emotionally extreme, polar, and emotionally charged material. This qualitative study adopts a conceptual and analytical research design-based on the conceptual dimensions of optimization models (predictive engagement metrics, behavioral profiling, dynamic feedback loops, and automated collaborative filtering are defined, and their systemic operation is correlated with Pakistan's very fragmented sociopolitical, sectarian, and hyper-partisan environment. This study can be seen as a tool for analyzing and breaking down the exact ways in which platforms like Facebook, TikTok, and X (formerly Twitter) facilitate epistemic isolation and create algorithmic religious authority. The study reveals the mechanism through which algorithmic bias creates localized vulnerabilities and sets in motion a process of failing to recognize non-violent political extremism and then speeds up processes of violent radicalization, particularly targeting young people. This paper recommends transparent algorithmic governance, practical policy options, laws, and engineering mitigations that could foster a counter-productive process without contravening core digital freedoms in transitional democracies.