People — not computers — should be deciding if my family can rent a home

My wife and I are preparing for a baby, and like any parents trying to get their lives in order, we need a place to bring our child home. 

During various rental application processes, I expected to provide proof of income, answer questions and follow application requirements. I did not expect to find myself questioning whether the people responsible for reviewing our information understood the process they were requiring us to use.

At one apartment community in San Diego County, we received an approval email and signed lease documents. Yet income verification continued to be an obstacle. We supplied documentation. We followed up. 

An employee acknowledged a calculation error, and our housing navigator questioned why verification remained unresolved. 

Having encountered application problems at another local property, I began paying closer attention to the distance between what we were being told and what was actually happening. 

An approval should have meant we could move forward. Instead, we were still trying to understand what was wrong and who could fix it. After all, I had spent eight years as a licensed mortgage banker, so reading W2s and 1099s, calculating income, auditing files, processing loans and supporting underwriting were part of my work experience. 

When the income calculations didn’t make sense, I had a professional reason to question them. I couldn’t inspect the software the apartment community was using, but I knew how to read our income documents and how to check the calculations myself. 

If someone with my background couldn’t get a clear resolution, what happens to an applicant who doesn’t understand how all this works?

That’s where my concern about artificial intelligence begins.

I’m Google AI-certified, and I use technology every day. I see its value. My concern is not that the technology should disappear. It is that a company should never let a system become the final answer for a rental decision when the evidence says otherwise. 

When a result conflicts with the information in front of an employee, someone needs both the authority and the responsibility to examine it, explain it and correct it. 

Otherwise, what purpose does the employee serve? Relinquishing our faculties to software leaves the person affected trying to reason with people who have stopped exercising their own judgment — their God-given ability to think — in favor of a machine.

As I looked beyond our experience, I found reasons to take that concern seriously. 

In 2022 the Consumer Financial Protection Bureau reported more than 26,700 complaints by renters, including those who faced inaccurate screening information and lengthy, confusing correction processes. 

That does not establish what caused our problems, but it shows that the difficulty of getting errors corrected extends beyond our application. 

In a landmark national class-action lawsuit, Louis v. SafeRent, Black applicants using housing vouchers alleged SafeRent’s algorithmic screening system disadvantaged Black and Latino applicants through its treatment of credit information and housing assistance data. 

The case settled in 2024 for $2.275 million without an admission of wrongdoing. 

Consider what reaching a courtroom requires of someone still trying to find a home. The ability to keep fighting a decision depends partly on what a person has left to fight with. 

Someone with savings may be able to absorb another application fee or a delay. Someone already stretched thin may not. 

Where financial disadvantage overlaps with racial inequality, an unresolved screening error can reinforce barriers that already exist. A family’s private struggle becomes a social problem when institutions control access to housing without providing an effective way to challenge data mistakes.

And who answers for these mistakes? The owners or managers of a property can point to the software provider. The provider can point back to the company making the housing decision. And its staff can point to company policy. 

Each explanation sends the applicant somewhere else. Follow that reasoning long enough and accountability seems to disappear, despite people having made decisions at every step. 

In their eyes, it can function like a liability shield, whether or not it offers any actual legal protection. Everyone can explain their limited role. But applicants are still left without a resolution.

I’d rather people be empowered to examine the evidence, explain the disputed result and correct an error before a housing opportunity disappears. 

Companies adopting these systems should train their staff and provide an alternative when the technology fails. Verifying income and preventing fraud are reasonable goals, but accuracy and accountability need to remain part of achieving that. 

Because a correction that arrives after a family has lost an apartment is hardly a workable solution.