Examining Racial Stereotypes in YouTube Autocomplete Suggestions

2024-10-04

arXiv.org

AI Summary

This study investigates how YouTube’s search auto-complete system systematically reinforces racial stereotypes and discrimination. Employing a novel five-dimensional sociocultural framework—encompassing appearance, competence, culture, social equity, and behavioral norms—the research introduces the concept of “aggregated discrimination” to characterize the algorithmic compounding of multiple biases against racial minorities (Asian, Black, Latinx, and White groups). Combining algorithmic output auditing with Critical Discourse Analysis (CDA), the study empirically demonstrates that auto-complete intensifies racial othering, exposes intergroup tensions, and perpetuates structural inequities. Findings reveal systematic bias in query suggestions, disproportionately associating minority groups with negative or reductive attributes. The work contributes both a practical bias-detection toolkit for platform governance and concrete policy intervention pathways. It constitutes the first empirical study to systematically deconstruct racialized auto-complete mechanisms on a major video-sharing platform.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Search and Optimization: Algorithm Configuration
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• Responsible Web: Algorithmic accountability and transparency on the web
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

Autocomplete is a popular search feature that predicts queries based on user input and guides users to a set of potentially relevant suggestions. In this study, we examine how YouTube autocompletes serve as an information source for users exploring information about race. We perform an algorithm output audit of autocomplete suggestions for input queries about four racial groups and examine the stereotypes they embody. Using critical discourse analysis, we identify five major sociocultural contexts in which racial biases manifest -- Appearance, Ability, Culture, Social Equity, and Manner. Our results show evidence of aggregated discrimination and interracial tensions in the autocompletes we collected and highlight their potential risks in othering racial minorities. We call for urgent innovations in content moderation policy design and enforcement to address these biases in search outputs.

Problem

• Analyzing racial stereotypes in YouTube autocomplete
• Identifying biases in autocomplete suggestions by race
• Highlighting risks of discrimination in search outputs

Innovation

• Algorithm output audit
• Critical discourse analysis
• Content moderation policy