Masterclass: Innovation Meets Regulation: AI, Real Estate, and Fair Housing Laws

Kathleen Lappe (CEO, Direct Offer), Liz Sturrock (Chief of MLS and Innovation, Miami MLS), and Nathan Brannen (CPO, Restb.ai) participated in a panel moderated by Audrey Whittington (VP/Head of Sales & Partnerships, Local Logic) to discuss the rise of generative AI and their impact on real estate, President Biden’s Executive Order on AI, and how to navigate those new frontiers to ensure they align with Fair Housing rules.

In this Masterclass, you’ll learn more about:

• Understanding Generative AI
  • Explore the latest advancements in AI technologies and their transformative role in real estate.
• Federal AI Compliance
  • Learn about the Executive Order’s implications for AI applications in real estate.
• AI’s Impact on Fair Housing
  • Analyze how generative AI aligns with the Fair Housing Act under the new federal guidelines.

Watch our latest Masterclass on “Innovation Meets Regulation: AI, Real Estate, and Fair Housing Laws”:

Understanding Generative AI

How can the real estate industry leverage AI?

A lot of what the real estate industry does daily is influenced by AI. Our panelists share how their companies harness AI to empower agents and homebuyers through content generation, imagery enhancements, and large language models (LLM) applications. 

For example,

• Miami Realtors uses AI to assist its realtors on the ground, particularly with content generation — whether that is for images (e.g. virtual staging and decluttering), blog posts, or social media posts.
• At the core of restb.ai is AI, which automates and structures imagery data at scale to better understand it.
• Direct Offer focuses on LLMs to create multilingual overviews of properties, ensure cohesive intent in translation models, and create engaging experiences for consumers and agents.

What role can AI play in generating residential leads?

Using AI to crawl massive databases, agents and brokers can gain further insights into consumer behavior and adapt their approaches accordingly to attract more clients.

Collecting and tracking data is not new. We’ve tracked what people make, how much their house is worth, whether they bought or sold a home recently, whether they applied for a mortgage, and whether they’d be perfect to sell to upgrade their home.

Today, our ability to track that information has increased in scale and specificity, allowing us to meet the needs of consumers more effectively. For example, beyond what’s inside the four walls of a home, the environment also plays a significant role in influencing the decision to buy. With a focus on the neighborhood, Local Logic’s location intelligence solutions help agents better understand what is meaningful to homebuyers in terms of where they live.

What are some challenges of AI — and how do we protect ourselves?

Maintaining data integrity and information accuracy

A major concern with AI is how it learns — what kind of information it receives and how it interprets it, as well as how much of it is generated for users. Therefore, people need to take responsibility for the AI-generated content they publish, by reviewing its accuracy and verifying its sources. 

This type of content, whether it’s for neighborhood descriptions or a guide on mold remediation, should primarily serve as a good starting point, but still be fact-checked before publication. AI is not a lazy solution. Despite its efficiency, AI still needs a final human touch.

Protecting asset ownership

A related concern is how AI affects copyrights and ownership of listings, MLS data, and photos. There is a balance between having access to data and rewarding those who own it. 

The MLS has done tremendous work to ensure that the data it provides is of the highest quality. It is certainly worrying that these new technologies could simply crawl the Internet and return information without any compensation. If you don’t have any regulations or incentives to motivate the MLS to standardize, collect, and make this information available, you end up with data that isn’t comparable in quality.

To navigate these unknown waters, the industry needs to be more regulated, as well as more transparent about who owns the data that’s being augmented, and what its origins are, which will likely lead to more data mapping and tracking.

Federal AI Compliance

What is Biden’s Executive Order on AI?

On October 30, 2023, President Biden issued an Executive Order addressing the safe, secure, and trustworthy development and use of AI. Key points include:

• Content Authentication: The Department of Commerce will guide content authentication and watermarking to label AI-generated content clearly.
• Skills Development: The federal government aims to support programs fostering AI-related skills among Americans, ensuring readiness for the AI era.
• Algorithm Assessment: Draft rules propose that federal agencies evaluate AI systems in use, particularly in law enforcement and healthcare, for potential harm to citizens.

President Biden’s order aims to establish safeguards and promote responsible AI development within the U.S. government.

What does it mean for the real estate industry?

An important point Biden raised in his recent AI Executive Order is algorithmic discrimination. What happens when historical data impacted by discriminatory practices, like redlining, is fed into models to predict future prices? 

There’s no one-size-fits-all solution. It’s important to handle both scenarios, i.e. what happens when a piece of data is wrong and what happens when a piece of data is right. Even if the AI is using the data correctly, understanding why it is being used is still necessary to ensure that biases aren’t perpetuated.

For example, a growing number of landlords are using AI to screen prospective tenants. The prevalence of inaccurate, outdated, or misleading information in the reports they use increases housing costs and barriers, especially for people of color. To remedy such situations, Biden’s Executive Order emphasizes transparency, responsiveness, and ethics when dealing with AI.

AI’s Impact on Fair Housing

NAR’s Federal Technology Policy Committee focuses on DEI and Fair Housing recommendations. Committee members are aware of incorrect appraisals delivered to African-American families and African-American-owned homes as well as skewed lending practices.

Algorithms were used to build all of that. The real estate industry needs to do better. There must be tests in place to ensure fair appraisals, fair AVMs, and fair lending. And these algorithms will need to be continuously backtested as they are updated.

How can AI contribute to Fair Housing practices?

Guarding against the ten forbidden words

Miami Realtors addresses Fair Housing with AI through the products they have partnered with. With the help of these vendors, the MLS ensured that the ‘ten words that you cannot say on television’ (i.e. words that will get you dinged and tested for Fair Housing) don’t appear in any published material, such as listing descriptions. While their members are still liable for everything that they post, the MLS wanted to provide them with some guardrails and base protection.

Phasing out discriminatory practices

The goal of Fair Housing is to prevent any type of discrimination, whether it be against a person’s race, sex, or religion — anything that shouldn’t be considered when renting or buying a house. It may manifest itself in simple things, such as not using one of those ten words, or in more tangible things, like valuing a home for a quarter less than it should have been.

AI can be used to identify and address inequalities in housing opportunities by analyzing housing patterns and trends. The use of new technologies could help prevent or minimize the impact of historical injustices, such as discrimination against people of color from purchasing a home.

Addressing disability with tech-driven solutions

People with visual impairments are unable to interact with a property the same way as those who can view photos, which is a big limiting factor to equal access.

Direct Offer aims to break down barriers to homeownership by making it easier for people with disabilities and able-bodied people alike to understand real estate listings. Its technology (e.g. multi-language audio tours) allows anyone to browse real estate listings, communicate with agents, and become a homeowner.

With restb.ai, all the images on a property can be automatically captioned, and listing descriptions are populated with descriptive text by partnering with MLSs. Home seekers can have those photos described so that they can understand and interact with each property in a way that they weren’t able to before.

Miami Realtors analyzed all of its websites to ensure that they were compliant. The MLS partnered with companies to make its websites as accessible as possible to both consumers and members, with initiatives like close-captioned classes, voice-activated webpages, resizeable text, and descriptive alt tags for photos.

Is AI the ultimate be-all and end-all for appraisals?

Are desktop appraisals phasing out appraisers? Not quite — the industry must identify when a full appraisal needs to be conducted on-site and when a desktop appraisal is sufficient.

Technology can, for instance, reduce the risk of manual errors in appraisals, as well as remove subjectivity from the condition of properties and comparables. 

However, as Greg Robertson says, “Zillow can’t smell the cat.” A desktop appraisal can’t determine whether a basement is musty. Those kinds of things are hard to tell until you’re on site. That’s why you need an appraiser to go beyond a heat camera and a LIDAR measurement. Humans are still needed for that.

In the end, a human and AI working together will produce the best results. Technology will always have a human component, as the former requires a higher intellectual guardrail — which is the point of the AI legislation.