Stanford-designed AI is helping this Bay Area county erase racist property records from the past

Dustin Dorsey Image
Friday, October 18, 2024
How AI is helping this Bay Area county erase racist property records
Santa Clara County is utilizing Stanford University-developed AI to redact thousands of racist property records from the system.

SANTA CLARA COUNTY, Calif. (KGO) -- Santa Clara County is utilizing Stanford University-developed artificial intelligence to redact thousands of racist property records from the system.

Racial covenants were utilized in the early 1900s to require property owners to follow certain rules including who can or cannot live there.

Even though they are not being enforced, these agreements still exist in documents to this day.

Stanford Law Professor Daniel Ho couldn't believe one particular clause in the documents when buying a home in Palo Alto.

"That specified that, 'the property shall not be used or occupied by any person of African, Japanese, Chinese or Mongolian descent, except for in the capacity as a servant to a White person,'" Ho said.

It's known as a racial covenant: a now unconstitutional agreement put in place by property owners, as recent as the early 1900s, that the purchaser had to abide by.

MORE: Parents sue school in Massachusetts after son punished for using AI on paper

In 2021, California required all 58 counties to proactively go through these records and remove any racial covenants that still persist in property records.

But in Santa Clara County, it wasn't that simple.

"We have approximately 24 million documents, or 84 million pages," Santa Clara Co. Asst. Clerk Recorder Louis Chiaramonte said.

"Yeah, that would be near impossible for a human to have to go through all that," ABC7 News reporter Dustin Dorsey said.

"Absolutely," Chiaramonte replied.

Having felt the impacts of racial covenants personally, Ho utilized his team at Stanford's Regulation, Evaluation and Governance Lab to assist the county.

MORE: Congressional candidate Sam Liccardo is using AI to speak to voters

RegLab used AI machine learning technology to remove the human element of this work.

The AI tool was designed to search through documents from 1902 to 1980 - the time period during which most of these racist covenants were created.

"That was able to sift through 5.2 million deed records to be able to find racial covenants amongst a lot of language," Ho said. "And that resulted in around 7,500 deeds (with racial covenants)."

It saved years worth of work - around 86,000 hours total.

Most importantly, it allowed for these harmful covenants to be removed from record.

"I would say there's probably a multitude of reasons at that time that they had it, which are all reasons that boggles my mind," Chiaramonte said. "Ultimately, it was correcting something that happened that was incorrect to have even happened in the first place."

Now Streaming 24/7 Click Here
Copyright © 2024 KGO-TV. All Rights Reserved.