Assessing the Impact of AI-Driven News Personalization on Public Trust

Thesis Abstract

The rapid integration of artificial intelligence (AI) technologies into news dissemination processes has fundamentally transformed the landscape of media consumption, raising critical questions about the implications of AI-driven news personalization on public trust in media institutions. This study investigates how AI algorithms that customize news content influence users’ perceptions of credibility, impartiality, and overall trust toward news providers. The increasing reliance on algorithmic curation, especially through social media platforms and online news aggregators, necessitates a comprehensive understanding of its social consequences, as concerns about information echo chambers, misinformation, and bias persist. The primary aim of this research is to assess the impact of AI-powered news personalization on public trust in media, with specific objectives to (1) evaluate the perception of credibility of personalized news feeds, (2) analyze the relationship between news personalization and perceived impartiality, and (3) determine differential trust levels among diverse demographic groups. To achieve these objectives, the study adopts a mixed-methods approach, combining quantitative surveys with qualitative interviews to provide an in-depth understanding of users' perceptions and experiences. The quantitative component utilizes a cross-sectional survey targeting internet users aged 18-65 across urban and rural settings, with a sample size of 600 respondents drawn through stratified random sampling. The survey instrument incorporates validated scales measuring trust, credibility, and perceived bias, and the reliability of these instruments is confirmed via Cronbach's alpha coefficients exceeding 0.8. Data analysis involves descriptive statistics to profile respondents, followed by multiple regression analysis using SPSS to identify significant predictors of trust influenced by news personalization. Structural equation modeling (SEM) is employed to test the hypothesized relationships based on the Elaboration Likelihood Model (ELM) and the Trustworthiness Theory, which posit that personalized information influences source credibility and user engagement. The qualitative data from 20 in-depth interviews are analyzed thematically using NVivo, allowing triangulation of the quantitative results and exploration of underlying perceptions, attitudes, and experiences regarding AI-driven news feeds. It is anticipated that findings will reveal a complex relationship whereby news personalization positively enhances perceived relevance and engagement, thereby increasing trust among certain demographic groups; conversely, it may also foster perceived bias and skepticism, particularly among active social media users and younger audiences. The study expects to demonstrate that personalization's effect on trust is mediated by perceived credibility and impartiality, moderated by factors such as media literacy, prior trust levels, and platform type. This research contributes to scholarly understanding of the social implications of AI in media by extending existing trust frameworks to the context of automated news curation. It advances theoretical insights by testing the applicability of the ELM and Trustworthiness Theory in the digital era, specifically within AI-mediated news environments. Practically, the study offers evidence-based recommendations for media practitioners, policymakers, and platform developers to optimize AI algorithms in ways that enhance transparency, fairness, and audience trust. The study concludes that balancing personalization with transparent disclosures and ethical standards is essential to fostering sustainable trust in AI-driven news ecosystems. It advocates for increased media literacy initiatives and regulatory frameworks to mitigate potential biases and misinformation. Future research directions include longitudinal studies to examine changes over time and the impact of emerging AI technologies on trust dynamics across diverse media landscapes.