AI Uncovers Racial Restrictions in Millions of California Property Records

AI Tackles Historical Housing Discrimination in California

In a groundbreaking collaboration, Stanford University's Regulation, Evaluation, and Governance Lab (RegLab) has partnered with Santa Clara County to leverage artificial intelligence in identifying and mapping racially restrictive covenants in property deeds. This initiative comes in response to a 2021 California law requiring counties to identify and redact discriminatory language from property records 1

1

The Challenge of Racial Covenants

Despite being constitutionally unenforceable since 1948, racially restrictive covenants continue to exist in deed records across the United States. These discriminatory clauses, which once prohibited property ownership or occupancy based on race, present a significant challenge for counties tasked with their removal 2

2

AI-Powered Solution

To address the daunting task of reviewing millions of documents, the Stanford team trained a state-of-the-art open language model to detect racial covenants with near-perfect accuracy. This innovative approach is estimated to save 86,500 person-hours and cost less than 2% of comparable proprietary models 1

1

Mapping Historical Discrimination

The researchers went beyond identification, developing a method to geolocate properties with racial covenants by cross-referencing historical maps. This process revealed striking insights into the evolution of housing discrimination in Santa Clara County:

1. Approximately one in four properties in the county were subject to racial covenants as of 1950.
2. Just 10 developers were responsible for a third of the identified covenants.
3. African Americans and Asian Americans were excluded at similar rates, despite significant population differences 2

   2

   .

Related Stories

AI Rental Screening Algorithm Settles Discrimination Lawsuit for $2.2 Million

AI hiring tools show racial bias against Black and Asian applicants, Stanford study reveals

AI Chatbots in Hiring: Uncovering Subtle Biases in Race and Caste

California Proposes New AI Regulation Framework Amid Growing Concerns

Implications and Future Applications

The success of this project demonstrates the potential for AI to assist in addressing complex societal issues. "We believe this is a compelling illustration of an academic-government collaboration to make this kind of legislative mandate much easier to achieve and to shine a light on historical patterns of housing discrimination," said Mirac Suzgun, a JD/Ph.D. student involved in the project 1

1

Broader Impact

By making their model publicly available, the Stanford team aims to enable other jurisdictions to efficiently identify, redact, and develop historical registers of racial covenants. This approach not only saves time and resources but also provides valuable insights into the historical patterns of housing discrimination 1

1

As communities across the United States grapple with the legacy of racial discrimination in housing, this AI-powered approach offers a promising tool for uncovering and addressing historical injustices while paving the way for more inclusive communities in the future.