/>
X
Innovation

How a real estate company is using AI to find and remove racist language from property documents

John L. Scott Real Estate is using Amazon Web Services technology to tackle the tough challenge of scrubbing racist language from property title documents.
Written by Stephanie Condon, Senior Writer on
using-pc
Image: Getty Images/fizkes

For too long, racist practices have plagued the real estate industry, and one glaring example of this is the racially restrictive covenants that still exist in countless property deeds in the US. 

These restricted who could buy, sell, lease, or occupy a property based on race. Even though the Fair Housing Act of 1968 ensured that these racist covenants are no longer enforceable, the ugly language first written into property deeds is often left there. It can be hard to find and even harder to officially remove. It's a problem that will take a long time and many parties to solve. 

Seattle-founded real estate firm John L. Scott is working with Amazon Web Services to build an intelligent document-processing program that could help homeowners easily identify racist covenants in their property title documents. Called the Driving Change program, it also simplifies the arduous process of getting property title documents legally changed. 

Without a tool like this, a homeowner would have to read their documents, find the language, figure out how the law with respect to amending property documents works, focus on how to get it notarized, and how to get it to the county recorder's office and so on, Phil McBride, operating officer for John L. Scott, told ZDNET. 

"The normal homeowner – who really does care – doesn't have the tenacity or time or expertise to do all that," he says. 

Also: The White House passes an AI 'Bill of Rights' that attempts to put your concerns at ease

The problem is hard to tackle manually, even when there are whole teams dedicated to the process. The University of Washington has been working to identify racist language in Washington state's property deeds for two decades and currently has a team of 650 volunteers reading deeds online. They've documented about 65,000 deeds, but there are literally millions of properties in the state that still need to be researched.  

"The only way we're going to get there is with technology," McBride says. 

John L. Scott built a website, powered by AWS, to help identify property titles that should be modified. A user types in their address, and if their property deed is digitized, the site uses Amazon Textract and Amazon Comprehend – leveraging computer vision and optical character recognition – to extract data from the documents and analyze it. 

If racist language is identified and the homeowner wants to change the document, the website will then provide them with the appropriate document to submit to their county recorder's office. It will auto-populate fields with information about the property, such as the parcel number and an abbreviated legal description. Then, the site prompts the homeowner to make an appointment with an online notary to finalize the change. 

"This is something we think we can scale," McBride says. 

Before the technology can be deployed, however, there are a couple remaining challenges to overcome. 

The first is technical. Only about 60% of the state's property deeds are digitized. And while Amazon's AI tools can analyze handwritten documents, there are other formats, such as microfiche, that present a challenge to the system. McBride said Amazon is helping John L. Scott address that issue. 

Then there are legal hurdles. Washington state's county recorders must determine how to proceed if there are multiple copies of a document at play. For instance, if a homeowner wants to modify their deed digitally, but a physical document or a microfiche version exists, which counts as the official document? 

While these issues are resolved, John L. Scott's website is in a "holding pattern", McBride says. He hopes the website will launch in Washington by early next year. 

Also: I tested out an AI art generator and here's what I learned

Meanwhile, if the company or others wanted to deploy this kind of tool in other states, more legal changes would probably be necessary. In Washington state, recently passed laws have made it easier to modify property titles with racist language. McBride himself helped shepherd the most recent bill through the state legislature, to help homeowners redact or officially repudiate racist language in their property titles.

Amazon Web Services, for its part, is highlighting the Driving Change project now in anticipation of other states following Washington's lead.

John L. Scott over the past two years has put more focus on the systemic racism within the industry, in part in response to its own history: CEO J. Lennox Scott only recently learned of the way his grandfather – the company's founder – interfered with one Black family's effort to buy a house in Seattle

"Our industry was very complacent and culpable when these restrictions were coming into play," McBride said, and as AWS notes in its blog post about the project: "Racial covenants are direct evidence of the real estate industry's complicity and complacency when it came to the government's racist policies of the past, including redlining." 

McBride says the Driving Change project isn't just about changing property deeds.

"It also brings up an opportunity to talk about things that are happening today that are preventing generational wealth for Black Americans," he said, citing issues like appraisal bias and racism within the credit-scoring system. "It opens up a conversation."

Editorial standards

Related

Using AI to help write code? The tech just isn't there yet, says a company that tried
Tech workers

Using AI to help write code? The tech just isn't there yet, says a company that tried

Five Asean nations ink pact to ease cross-border payments
world-connected.jpg

Five Asean nations ink pact to ease cross-border payments

This giant supercomputer dominates the rankings, and catching it will take a lot of work
shutterstock-388828003.jpg

This giant supercomputer dominates the rankings, and catching it will take a lot of work