AI in the Racial Wealth Gap: Deciding our Future (Full report)
AI in the Racial Wealth Gap: Insights (PDF)
Published June 2026
The racial wealth gap in the United States is one of the most persistent measures of economic inequality. Without safeguards, AI will widen this gap — not narrow it.
To better understand how AI systems affect individuals and shape the racial wealth gap, we conducted a literature and policy review alongside exploratory qualitative interviews with people of color. The interviews revealed six key themes:
AI systems compound barriers across domains.
Participants described how exclusion in one domain, such as housing, quickly cascaded into other areas of life. For example, a housing denial led some to take on additional credit card debt, which in turn negatively impacted their credit score, further closing off future opportunities. Rather than acting as discrete barriers, algorithmic harms are often interlocked, compounding financial and psychological stress.
“At the time, I was very stressed….it was terrible, because it was my livelihood, the hotel, it was getting so expensive, and I just didn’t know…how I’ll be able to sustain myself. I haven’t gotten my first paycheck yet, because when you start a new job, you usually don’t get the paycheck of the first cycle, you usually get it that following the cycle. So, that was like a month of me waiting for a paycheck…. I don’t come from the most privileged background, so I didn’t want to cause any financial stress on my parents. They’ve done so much already, and even then. I don’t want to put them in a bad situation and help me. Because I wanted to make the move.”
Algorithmic opacity leads to feelings of powerlessness.
A consistent theme was the absence of transparency. Interviewees frequently reported not knowing whether an algorithm had been used to screen them, what criteria had been applied, or how to contest an outcome. This opacity reinforced feelings of helplessness and disillusionment with the institutions with which they were engaging.
“…I think that the system inadvertently assumes that people of color are higher risk, and so they probably end up getting higher interest rates just because of that. I mean, that’s where I feel that I am, like, I…don’t understand why with, like, my credit score and history and all of that combined, it’s not enough to get a rate that’s, like, lower or on the average rate. It’s always on the high end, and that doesn’t make sense to me.”
Human oversight is a necessary protective factor.
While many participants experienced blanket rejections through automated systems, some recounted more equitable outcomes in settings where human decision-makers were involved, particularly when those decision-makers shared lived experiences with applicants. For example, one participant only secured housing after speaking directly with Black property managers who were able to look beyond a credit score and consider broader financial context.
“I’m used to a leasing agent, or somebody to talk to. Instead, there’s this bot named Amy…and I don’t want Amy, I want a person.”
Credit-based exclusions reinforce structural inequality.
Several participants had average or above-average credit histories with no delinquencies but were still denied housing or loans. Common credit model inputs, like student loan balances, triggered denials despite participants’ consistent repayment.
“You know, you try to play the game, you do the loans, you do the payments, you build credit, you apply for a bunch of credit cards so that you have, like, more credit available to you, and that, you know, in some imaginary land makes you look like a better person to…[let borrow money]. I’m just seeing…from my perspective, I’ve done everything that I could…[m]y credit score is, like, [c]lose to 800…But, you know, I’m finding that that’s not really worth anything…Everything that we’re supposedly doing right that’s supposed to help us get to where we need to be. You do it, and… [it’s] still not enough.”
“Ghost postings” and automated rejection in employment create confusion and frustration.
In the job market, participants described a demoralizing landscape dominated by AI tools and deceptive practices. Many reported applying to roles that turned out to be “ghost postings”—jobs that were already filled or never intended to be filled. Combined with algorithmic résumé screening, applicants were left uncertain not only about whether they were fairly evaluated, but whether the opportunity even existed.
“[The company] ha[s] an applicant-tracking system that could completely disregard your resume if it doesn’t have a certain skill word in there…they don’t even send it to the actual hiring manager to review it and read it.”
Overuse of algorithms erodes hope and mobility.
While participants entered systems in pursuit of economic mobility by applying for housing, credit, or jobs, many left feeling worse off. Debt accumulation, loss of time, and lack of recourse led to a sense of futility. Several articulated that AI systems, rather than expanding opportunity, often functioned to foreclose it.
“It feels very… misleading in a way, because we’re told, like, hey, if you have good credit, if you follow the rules, if you do everything this way, you go to school, you do this, you check all the boxes, you should be able to afford a home, or get a small loan to get a piece of land. So it felt like everything that… I was told to work towards, it’s just not enough. That’s, I guess, the feeling of… that’s upsetting.”