A Cross-Temporal Analysis of Redlining Language and Modern Tenant-Screening Algorithms

This paper examined how artificial intelligence was used in tenant screening. Previous research focused on the quantitative and technological aspects of AI in housing, particularly bias and socioeconomic impact. However, no cross-temporal studies have been conducted to analyze similarities in the language used to justify decisions between these two eras. This left a gap in understanding how the continued use of the language itself may influence inequality. To address this, the study examined whether modern screening systems repeat patterns of “risk” justification through a cross-temporal, mixed content analysis, sampling phrases and rhetorics from 16 Home Owner’s Loan Corporation documents and 12 modern tenant-screening papers. By coding 140 lines of data into 5 main categories that best described how they function in the texts, the study identified how decisions, discriminatory or not, have been justified across time periods. The findings of this paper indicate that all main categories of justifications consistently appear in both historical and modern datasets, suggesting that modern systems were still influenced by historically discriminatory ideas. Ultimately, this study provided a new perspective on AI use in the modern era, showing that the logic of prejudice can manifest through qualitative language, even when quantitative outcomes or technical information contradict it. Keywords: Redlining, Risk Rhetoric, Algorithmic Bias