Algorithmic discrimination: a grounded conceptualization

Matthew Bui

doi:10.1080/1369118X.2025.2516544

2025-06-19

Matthew Bui *

Charlton D. McIlwain

Seyi Olojo

Ho-Chun Herbert Chang

DOI：10.1080/1369118X.2025.2516544

原文链接

原文求助

摘要

En 中文

This paper extends the literature on algorithmic discrimination to analyze the racialized harms and outcomes of online targeting. In addition to providing a genealogy, we argue that targeted ads are apparatuses designed to discriminate: to find patterns based on past racial trends; and selectively expose communities to resources (while excluding others), under promises of personalization and optimization. Rooted in a sociohistorical and infrastructural approach to examining algorithmic ethics and impacts, our approach establishesalgorithmic discriminationas an analytic tool for drawing increased attention to the disparate outcomes of online targeting. Importantly, our approach expands on existing frameworks by conceptualizing algorithmic discrimination through the everyday lived experiences of individuals and communities disproportionately impacted by online targeting. We contend that a grounded conceptualization, coupled with critical race scholarship to historicize and contextualize these experiences, spotlights the complex and nuanced ways online targeting exacerbates the advertising industry’s fraught, racialized history of commercial data collection and analytics as well as how it expands larger systems of automated racial classification, surveillance, and inequality. Moreover, a grounded and historically situated conceptualization is especially effective for auditing and accounting for community – and systems-level harms perpetuated by algorithmic discrimination because it can foster more holistic and effective interventions.

Keyword:

AI ethics

algorithmic discrimination

algorithmic fairness

racial inequality

targeted ads