This is the first edition of capsule summaries of recent research on Fair Housing and related subjects.
Artificial Intelligence (AI) and Discrimination.
Association of Historic Redlining and Present-Day Health in Baltimore.
Obstacles to Inclusion and Threats to Civil Rights: The Many Negative Social Experiences of Service Dog Partners in the U.S.
Fighting Discrimination Against Mothers of Children with Disabilities.
Relationship Between Residential Racial & Economic Segregation and Main Causes of Death in U.S. Counties.
Diversifying but not Integrating: Entropic Measures of Local Segregation in Philadelphia.
Black Exclusion, Racial Segregation, and Spatial Integration in U.S. Municipalities, 1990 - 2020.
Interesting Recent Research on Housing Segregation: Part 1: Wealth, Schools, Transportation, Noise.
"Artificial Intelligence (AI) and Discrimination"
"Artificial Intelligence and the Discrimination Injury," by Andrew D., (March 14, 2025). UCLA School of Law, Public Law Research Paper No. 25-10, 78 Florida Law Review. (forthcoming 2026), Available at SSRN: https://ssrn.com/abstract=5179224 or http://dx.doi.org/10.2139/ssrn.5179224.
For a decade, scholars have debated whether discrimination involving artificial intelligence (AI) can be captured by existing discrimination laws. This article argues that the challenge that artificial intelligence poses for discrimination law stems not from the specifics of any statute, but from the very conceptual framework of discrimination law. Discrimination today is a species of tort, concerned with rectifying individual injuries, rather than a law aimed at broadly improving social or economic equality. As a result, the doctrine centers blameworthiness and individualized notions of injury. But it is also a strange sort of tort that does not clearly define its injury. Defining the discrimination harm is difficult and contested. As a result, the doctrine skips over the injury question and treats a discrimination claim as a process question about whether a defendant acted properly in a single decision-making event. This tort-with-unclear-injury formulation effectively merges the questions of injury and liability: If a defendant did not act improperly, then no liability attaches because a discrimination event did not occur. Injury is tied to the single decision event and there is no room for recognizing discrimination injury without liability. AI works by using algorithms (i.e., instructions for computers) to process and identify patterns in large amounts of data (“training”), and then use those patterns to make predictions or decisions when given new information.
Researchers and technologists have repeatedly demonstrated that algorithmic systems can produce discriminatory outputs. Sometimes, this is a result of training on unrepresentative data. In other cases, an algorithm will find and replicate hidden patterns of human discrimination it finds in the training data. A 2021 analysis by The Markup of mortgage lenders who used underwriting algorithms found that the lenders were far more likely to reject applicants of color than white applicants: 40% more likely for Latino or Hispanic Americans, 50% more likely for Asian Americans and Pacific Islanders, 70% more likely for Native Americans, and 80% more likely for Black Americans.
This formulation directly affects regulation of AI discrimination for two reasons: First, AI decision-making is distributed; it is a combination of software development, its configuration, and its application, all of which are completed at different times and usually by different parties. This means that the mental model of a single decision and decisionmaker breaks down in this context. Second, the process-based injury is fundamentally at odds with the existence of “discriminatory” technology as a concept. While we can easily conceive of discriminatory AI as a colloquial matter, if there is legally no discrimination event until the technology is used in an improper way, then the technology cannot be considered discriminatory until it is improperly used.
The analysis leads to two ultimate conclusions. First, while the applicability of disparate impact law to AI is unknown, as no court has addressed the question head-on, liability will depend in large part on the degree to which a court is willing to hold a decisionmaker (e.g. and employer, lender, or landlord) liable for using a discriminatory technology without adequate attention to the effects, for a failure to either comparison shop or fix the AI. Given the shape of the doctrine, the fact that the typical decisionmaker is not tech savvy, and that they likely purchased the technology on the promise of it being non- discriminatory, whether a court would find such liability is an open question. Second, discrimination law cannot be used to create incentives or penalties for the people best able to address the problem of discriminatory AI - the developers themselves. This Article therefore argues for supplementing discrimination law with the application of a combination of consumer protection, product safety, and products liability - all legal doctrines meant to address the distribution of harmful products on the open market, and all better suited to directly addressing the products that create discriminatory harms.
Read the September 13, 2024 Brookings article.
Read the August 25, 2021 The Markup article.
(Image by storyset on Freepik.com.)
Labels: AI discrimination, applicants of color, artificial intelligence, discrimination law, discriminatory terms, mortgage lenders
"Association of Historic Redlining and Present-Day Health in Baltimore"
"Association of historic redlining and present-day health in Baltimore," by Huang SJ, & Sehgal NJ (2022). PLoS ONE 17(1): e0261028. https://doi.org/10.1371/journal.pone.0261028.
In the 1930s, the government-sponsored agency Home Owners’ Loan Corporation (HOLC) - created as part of the New Deal in 1933 - categorized neighborhoods by investment grade along racially discriminatory lines, a process known as redlining. Although other authors have found associations between HOLC categories and current impacts on racial segregation, analysis of current health impacts rarely use these maps.
A study released in 2018 found that 74% of neighborhoods that HOLC graded as high-risk or "hazardous" are low-to-moderate income neighborhoods today, while 64% of the neighborhoods graded "hazardous" are minority neighborhoods today. "It's as if some of these places have been trapped in the past, locking neighborhoods into concentrated poverty," said Jason Richardson, director of research at the National Community Reinvestment Corporation (NCRC), a consumer advocacy group.
Fifty-four present-day planning board-defined community statistical areas are assigned historical HOLC categories by area predominance. Categories are red (“hazardous”), yellow (”definitely declining”) with blue/green (“still desirable”/”best”) as the reference category. Community statistical area life expectancy is regressed against HOLC category, controlling for median household income and proportion of African American residents.
Red categorization is associated with 4.01 year reduction (95% CI: 1.47, 6.55) and yellow categorization is associated with 5.36 year reduction (95% CI: 3.02, 7.69) in community statistical area life expectancy at baseline. When controlling for median household income and proportion of African American residents, red is associated with 5.23 year reduction (95% CI: 3.49, 6.98) and yellow with 4.93 year reduction (95% CI: 3.22, 6.23).
The primary policy implication is that discriminatory public or social policy - whether or not such a policy is intentionally discriminatory - in a realm such as housing can potentially have long-lasting, disparate, and large impacts on health. Additionally, contemporary African American activists and organizations in the 1930s presciently recognized the discriminatory impacts of how HOLC made its policy decisions: members of marginalized and/or impacted communities should have control over every major policy-making process.
Since many of these issues are structural and historical, health interventions that do not focus on disparate health outcomes risk being “weighed down” by structural problems that predispose a population to worse health. If we take health equity and disparities research seriously, we should be examining the possibility of interventions that attempt to tackle some of these structural factors, including the possibility of reparations. What Link and Phelan propose as “fundamental causes” of health disparities - such as a lack of flexible resources of money, knowledge, social connections, and political power - may have even more fundamental causes rooted in power hierarchies such as racial, class, and gender subordination.
Results add support that historical redlining is associated with health today. Even time-limited urban changes can have long-lasting cumulative effects. Michaels and Rauch found that the differential collapse of Western Roman urbanization in Britain and France in the 6th century CE differentially impacted the spatial efficiency of urbanization even 1500 years later in the 21st century. Similarly, evidence supports that redlining still has cumulative impacts on various social factors today. Aaronson et al. found that living in close proximity on two sides of differently graded borders - as represented in the 1930s HOLC security maps - is strongly associated nation-wide with increased residential racial segregation from the 1930s to today. The effect on residential racial segregation was particularly strong from 1930-1970 and the effect size began to decrease after 1970. Aaronson et al. provide additional support that maps were likely drawn with race in mind: only areas marked category D had a primarily Black population in the 1930s. In addition to increased segregation, Aaronson et al. found support for reductions in home ownership, house values, and credit scores throughout the 20th century that are maintained even today nearly a century later. They also found evidence of “yellow-lining”: areas marked as category C also had disparate current outcomes when compared to higher rated areas. Using a different methodology, Appel & Nickerson found that redlined neighborhoods had lower home prices in 1990 compared to surrounding areas, and that these discriminatory effects remained even after nearly 60 years. The presence of these discriminatory effects can be compounded across time: Massey et al. found that Black residents of redlined neighborhoods face greater barriers to residential mobility than white residents that negatively impacts Black residents’ social and economic well-being.
"Obstacles to Inclusion and Threats to Civil Rights: The Many Negative Social Experiences of Service Dog Partners in the U.S."
"Obstacles to inclusion and threats to civil rights: An integrative review of the social experiences of service dog partners in the United States," Leighton SC, Hofer ME, Miller CA, Mehl MR, Walker TD, MacLean EL, et al. (2025) . PLoS ONE 20(3): e0313864. https://doi.org/10.1371/journal.pone.0313864. Editor: Laura Hannah Kelly, Public Library of Science, United Kingdom of Great Britain and Northern Ireland. Copyright: © 2025 Leighton et al. This is an open access article distributed under the terms of the Creative Commons Attribution License. Open Science Framework: https://osf.io/hj9nb.
Service dogs, trained to assist people with disabilities, are known to impact their human partners’ social experiences. While service dogs can act as a “social bridge,” facilitating greater social connection under certain circumstances, many service dog partners also encounter challenges in social settings because of the presence of their service dog – despite legal protections. Among the most common challenges reported are experiences of stigma, discrimination, and access or service denials.
This review sought to synthesize empirical, theoretical, and legal literature to understand better the social experiences reported by service dog partners in the United States, including (1) civil rights experiences; (2) experiences of stigma and discrimination; and (3) broader social experiences. Following database searches and article screening, a total of N = 43 articles met the eligibility criteria for inclusion. Analyses were conducted in two stages: first, synthesizing quantitative and qualitative findings to explore the magnitude of social experiences reported by empirical articles and second, narrative synthesis to integrate findings across all article types. Analyses identified three themes: Adverse Social Experiences, Contributing Factors, and Proposed Solutions. Overall, we found consistent reports of stigma, discrimination, and access denials for service dog handlers. Additionally, these adverse experiences may be more common for service dog partners with disabilities not externally visible (i.e., invisible disabilities such as diabetes or substantially limiting mental health conditions).
This integrative review highlights a pattern of social marginalization and stigmatization for some service dog partners, exacerbated by inadequate legal protection and widespread service dog fraud. They experienced social marginalization in the form of stigma, discrimination, and access denials. Moreover, service dog partners with invisible disabilities may be at higher risk for adverse social experiences. The lack of adequate legal protections at federal and state levels and the high prevalence of service dog fraud contribute to these issues. These findings have implications for the individual well-being of people with disabilities partnered with service dogs and highlight a need for collective efforts to increase inclusion and access.
In 1990, the Americans with Disabilities Act (ADA) was created to safeguard the civil rights of people with disabilities by establishing “clear, consistent, and enforceable standards". The ADA was born out of the disability advocacy movement, which aims to counter the ableism and historical marginalization that people with disabilities have long faced. It was enacted after decades of activism and advocacy by individuals with disabilities who fought to raise public awareness of the barriers they faced; these barriers included inaccessible environments, inequitable medical treatment, barriers to self-determination, and obstacles to economic participation. Ultimately, the ADA was signed into law on July 26, 1990 (42 U.S.C. § 12101). However, despite the legislative safeguards put in place by the ADA and related laws, people with disabilities in the United States continue to face various systemic barriers that impede their access to healthcare, education, employment, and community involvement.
One of the legal rights afforded to individuals with disabilities by the ADA is the right to be accompanied by a trained service animal, defined as a dog or miniature horse that is “individually trained to do work or perform tasks” directly related to the person’s disability and for the benefit of the individual (28 CFR § 36.104). In legal terms, service animals are akin to assistive technology like wheelchairs, prostheses, or hearing aids [8] and can be acquired through a professional organization or “owner-trained.” The legal right of a person with a disability to be accompanied by their service animal applies in most public spaces, regardless of any existing companion animal restrictions and provided that the animal is under the handler’s control. Under the ADA, employees of businesses may ask people with service animals (i.e., service animal partners) two questions to determine whether a dog or miniature horse is a service animal (1) “Is this a service animal required because of a disability?” and (2) “What work or task has the dog been trained to perform?” Service animals are not required to wear a vest or identification or to be “certified” as an indicator of legitimacy. However, public access rights for service animal partners do come with contingencies. For example, service animals may lawfully be denied entry in spaces where their presence would fundamentally alter the business or entity, such as sterile environments in a hospital or certain parts of a zoo wherein seeing or smelling a service animal could disrupt the resident animals (28 C.F.R. § 36.302). However, even in these contexts, the person with a disability must still be allowed to stay and use the facility without the dog present. It is also important to note that service animal regulations may not apply to certain entities, such as certain federal or religious organizations exempt from adhering to the ADA.
While service dog partnership was associated with increased quantity and quality of the human partners’ social interactions in some respects, concerns for the social well-being of service dog partners were also noted. Some service dog partners reported routinely encountering inconvenience, unwanted attention, and rude or poor behavior from others when with their service dogs in public settings. Moreover, there were numerous reports of service dog partners experiencing stigma, discrimination, and access denials in public settings – so much so that these are suggested to be defining experiences for members of an “assistance dog sub-culture” within the broader disability community. This sub-culture – the service dog community - offers empathy and essential support to its members who have faced various challenges, from microaggressions and intrusive questioning to outright access denials or service refusals. Beyond community support, legal cases have also arisen from access issues in various settings, with varying outcomes.
While it seems that most instances of stigma and discrimination towards service dog partnerships came from non-disabled people, this was not always the case. Stigma against service dogs can also exist within disabled communities themselves. For example, some Deaf and blind people may hold cultural stigmas against hearing or guide dog partnerships due to concerns that (1) the partnership might make them appear “disabled,” whereas not all members of these communities identify as disabled in the first place, and (2) service dog partnerships could potentially generate further barriers to equal societal participation for those individuals both with and without dogs.
(Image by prostooleh on Freepik.com).