On April 29, 2024, HUD released important guidance on the use of Artificial Intelligence (AI) in tenant selection decisions. The document highlights how even relying on seemingly objective tools may not provide enough context to arrive at appropriate tenant admission determinations.

Applicant screening is a vital part of the tenant selection process, but it is easy to lose sight of why. It is HUD’s intent to make its affordable housing opportunities available to a wide array of people across the country, regardless of their race, color, religion, sex, national origin, disability, or familial status. Still, operators of affordable housing programs must take steps to safeguard programs in a way that both prevents access to housing for people who pose a danger to others and generally provides a safe place to live.

In 2015, HUD affirmed that overly burdensome or restrictive screening policies may unintentionally create disparate impacts on people of certain protected classes. Facially neutral policies designed to select only applicants with pristine criminal backgrounds or financial histories can have a greater effect on entire classes of people who, like so many, are in critical need of an affordable home.

To streamline and systematize the selection process, many property management professionals have begun relying on background screening companies and software that make housing decisions based on data points like criminal history, credit scores, and eviction records. The prevailing thought is that a computer program that relies on specific criteria to choose suitable tenants is preferable to a person who may carry inherent biases.

However, it is important to recognize that AI systems, while designed to be objective, can also reflect and perpetuate existing biases present in their training data. AI programs fundamentally use machine learning to build databases and understand patterns in the data. While AI can identify patterns, it does not inherently understand the context behind those patterns.

For instance, if an AI program is taught that a recent misdemeanor is a barrier to housing, it may start to associate misdemeanors with other negative factors like poor credit ratings or eviction records. Without context, the AI may incorrectly infer that these factors are inexorably linked. AI systems may not account for the nuanced socio-political factors that contribute to these records, such as historical biases and systemic inequalities.

Consider a scenario where a person with a recent misdemeanor for fraud may have been compelled to falsify a check under the threat of physical harm from a domestic partner. Or a person with a bankruptcy due to extensive medical debt, or an eviction record stemming from job loss due to a new disability. In this scenario, these people with negative results on their screening history are perhaps the most in need of affordable housing options. These contexts are crucial for understanding an applicant’s true suitability for housing, and AI systems may overlook them if not properly designed and monitored.

All of this doesn’t mean you should outright reject AI in tenant screening. When selecting a company or product, make sure you’re able to customize and provide detailed information about the context of negative entries so you can align the process with your Tenant Selection Plan. Don’t purchase a product that will screen for unnecessary information, such as whether they’ve had multiple phone numbers or addresses in the last several years. If available, select a product that requires “human” attention to particular negative references. Most importantly, always review the entirety of the report with context in mind. Applicants have the right to appeal rejection decisions, and during their informal hearing, you may be presented with mitigating factors or with new information that can assure you that they are able to fulfill the terms of your lease.

Remember: you’re ultimately screening for suitability for tenancy. You’re not making a character judgment or looking for reasons to reject needy applicants. You’re simply fulfilling HUD’s goal to create and preserve decent, safe, and sanitary housing for Low-Income people in the US.

Share This