How AI Exposed 75 Years of Housing Discrimination Hidden in 5.2 Million Property Records

Author(s): MKWriteshere

Originally published on Towards AI.

A technical deep dive into Stanford’s fine-tuned language model that saved 86,500 hours of manual work and revealed shocking patterns of racial exclusion

AI uncovered what manual review couldn’t: systematic racial exclusion where one in four properties barred Black and Asian families, even when their populations were tiny. Technology finally made the invisible visible

(Non-Member Link)

Imagine discovering that one in four properties in your county was once subject to racially discriminatory restrictions.

Now, imagine these restrictions, despite being illegal for over 50 years, are still buried in millions of property records.

This was the reality facing Santa Clara County when California’s Assembly Bill 1466 required all 58 counties to find and remove racially restrictive covenants from their property records.

The challenge was massive. Santa Clara County had over 24 million property deed documents dating back to the 1850s. Manual review would take one person 160 years and cost over $22 million.

The solution? Artificial intelligence.

The Hidden Crisis: Understanding Racially Restrictive CovenantsBuilding the AI Solution: Technical Architecture and PerformanceShocking Discoveries: Geographic Patterns and Developer ResponsibilityThe True Scale: From 7,500 Deeds to 24,522 PropertiesImpact and Future: Open Source Tools for Legal Reform

Racially restrictive covenants were legal clauses written into property deeds. These clauses prohibited the sale, lease, or occupation of land based on race. They “ran with the land,” affecting not just current owners but… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI