Summary: The AI That Assumes You Exist

Rosa Mendoza is 71 years old. She has lived in the United States for thirty-one years. She raised three children here, two of whom graduated from college. She paid taxes using an Individual Taxpayer Identification Number for twenty-eight of those years. She cleaned houses, then office buildings, then worked in a restaurant kitchen until her knees gave out at 64. She has contributed to a country that does not, in the formal language of its benefit systems, know she exists.

The benefits navigation agent from Series 2, the AI that identifies every program a person qualifies for and helps them apply, gave Rosa an honest answer. Medicare requires forty quarters of covered employment with a Social Security number. Rosa has a taxpayer identification number, not a Social Security number. Social Security retirement benefits require the same qualifying record she does not have. The Affordable Care Act marketplace is closed to undocumented immigrants regardless of how long they have lived, worked, and paid taxes. The agent is not broken. The programs it navigates were designed for people with documentation Rosa does not have. What remains on the list: emergency Medicaid for acute care, community health center services on a sliding fee scale, and a small number of state-funded programs that do not require immigration status verification. The list is short. Rosa is not invisible to the healthcare system. She is visible as a patient who cannot pay. She is invisible as a citizen owed anything in return.

The article does not stop at Rosa. It names four other populations the AI ecosystem excludes through its assumptions.

The privacy architecture problem for undocumented elders: a benefits navigator that requires disclosure of immigration status to evaluate eligibility creates a record that, depending on the political environment and the platform’s data-sharing practices, can become a liability rather than a resource. Privacy-protective architectures that assess eligibility without requiring that disclosure are technically achievable using differential privacy techniques and secure computation. The system can tell Rosa what she may qualify for without creating a deportable record. This architecture is not standard. Putting documentation first was a design choice. A different design choice would put the person first.

The formerly incarcerated elder: roughly 200,000 state and federal prisoners are over 50, and the mass incarceration wave of the 1980s and 1990s is producing an aging cohort navigating elder care systems with records that close public housing, restrict benefit programs, and carry decades of institutional distrust that the AI ecosystem, designed for institutional comfort, does not account for. Benefits navigation that acknowledges record-based eligibility restrictions rather than presenting programs the person cannot access is a different product from what currently exists.

The LGBTQ+ elder: the personal AI that coordinates couples-based care, involves a partner in cognitive monitoring, and includes relationship status in care planning assumes a safe environment for those disclosures. Research documents that LGBTQ+ older adults in long-term care facilities report high rates of discrimination and the pressure to conceal identity and relationships. An AI that asks about the user’s partner in a facility where a same-sex couple has learned to introduce each other as friends is doing exactly what it was designed to do, in an environment where the design assumption is wrong. Separating the information the AI needs for care coordination from the information that creates institutional risk is a design choice that has not been made standard.

The elder with disability: roughly 46 percent of Americans over 75 have at least one disability. A person who is blind cannot use the visual scaffolding system from Series 5. A person who is deaf cannot use the speech-based daily check-in from Series 1. Voice-activated home controls do not work for a person who cannot speak. Accessible design as a standard rather than a retrofit produces a different product: an interface designed from the beginning to work for both sighted and blind users is not a concession, it is usually clearer and simpler for everyone.

The Indigenous elder: ambient home monitoring assumes a single-family dwelling with broadband. Many Indigenous elders live in multigenerational households on reservations where broadband penetration is among the lowest in the country. Indian Health Service, chronically underfunded at roughly one-third of national per-capita healthcare spending, cannot provide the clinical infrastructure that AI monitoring is designed to supplement. Tribal data sovereignty, the principle that data generated by tribal members on tribal land belongs to the tribe, is an emerging legal and ethical framework that most AI health companies have not engaged. An AI system that collects health data from Indigenous elders without tribal consent is not a neutral technology deployment.

Each exclusion named in the article is addressable with a specific design choice. Privacy architectures that do not require documentation disclosure. Benefits navigation that accounts for record-based restrictions. Interfaces that separate care information from identity risk. Accessibility built in from the first prototype. Tribal data partnerships that respect sovereignty. None of these are impossible. All of them are uncommon. The question is not whether the technology can accommodate the people it currently excludes. The question is whether the people building the technology consider them worth designing for.

Rosa Mendoza paid taxes for twenty-eight years. The agent that could help her most was designed for someone else. The design that would include her is a choice that has not been made.

Read the full article at BlueMirror.life.