Michael Higasi

When I began my internships at three eldercare facilities, I approached eldercare through a familiar assumption: older adults need more assistance, more companionship, and technologies designed especially for them. This assumption seemed reasonable. Aging can bring changes in mobility, health, and access to social support, while artificial intelligence can provide health monitoring, emergency alerts, and virtual companionship. However, after examining residents ' responses to a small set of questionnaires about A I-assisted care and dining services, I began to question whether providing more help was always the right starting point. The central issue was not simply what technology could do for older adults, but whether it allowed them to retain control over their own lives.

Over three months, I collected 30 questionnaires from three eldercare facilities, with 10 collected at each facility. Most were completed by residents, while a smaller number were completed by staff members. The questionnaires covered two topics: attitudes toward AI-assisted care and satisfaction with the facilities' dining services. Participation was voluntary, and the questionnaires did not collect respondents’ names or other identifying information. Because this was a small exploratory investigation, my purpose was not to rank the three facilities or generalize about all older adults. Instead, I wanted to identify patterns in how people living and working in different eldercare environments understood their needs.

The three facilities represented different models of eldercare. Taikang Community Shen Garden is a large suburban continuing-care community in Songjiang. According to its official website, it has approximately 2,100 senior-living units and combines residential, medical, and lifestyle services within a large community setting (Taikang Community, n.d.-b). Taikang Community Jing'an Mansion, by comparison, is an urban senior community located closer to central Shanghai. Its official description emphasizes multidisciplinary support, integrated medical care, rehabilitation, and cultural and recreational activities (Taikang Community, n.d.-a). Hengyu Caojiadu Senior Welfare Home is a publicly built and privately operated urban welfare institution. Government information describes it as an urban eldercare complex combining residential care, community services, day care, medical support, and home-based care; it has 499 registered beds (Jing'an District People's Government, 2026; Shanghai Civil Affairs Bureau, 2021).

These public descriptions do not tell us how every resident experiences each facility, but they frame the comparison: the three institutions differ in scale, location, physical setting, and service model, all of which may shape the expectations residents bring to their care.

My questions were also shaped by a concern I noticed in the language surrounding the eldercare industry. Products described as "elderly technology" frequently assume that all older users need simplified interfaces, constant monitoring, or assistance with ordinary decisions. The World Health Organization defines ageism as stereotypes, prejudice, and discrimination based on age and warns that it operates in institutions as well as in interpersonal relationships (World Health Organization, 2021). Researchers in human-computer interaction have observed a similar pattern in technology design. Vines et al. (2015) found that research has often presented aging primarily through frailty, dependency, and technological difficulty. This approach can encourage designers to build products for an imagined, uniform "elderly user" rather than for individuals with different abilities, preferences, and experiences.

Ageism can therefore appear even in services created with good intentions. A product may be designed to protect older adults while denying them the opportunity to decide what kind of protection they want. A simplified interface may help one person but frustrate another who already uses ordinary digital products confidently. A monitoring system may make one resident feel safe but make another feel that their privacy is being compromised . The problem is not that specialized support exists; it is that age alone is sometimes treated as sufficient evidence that everyone requires the same support.

Responses did not reveal one universal attitude toward AI. Some respondents indicated interest in functions such as fall detection, health monitoring, and faster connections to medical professionals . Others were hesitant about AI, placed greater importance on human service, or were willing to consider technology only under certain conditions, such as affordability and ease of use. Concerns included malfunction, inaccurate information, privacy, and reduced human interaction.

The most important pattern was not agreement but variation. Differences also appeared within individual facilities, although the small number of responses does not allow firm comparisons. This finding is consistent with Peek et al.'s (2014) systematic review of technology for aging in place. The review found that older adults' acceptance of technology depends on factors including perceived usefulness, cost, privacy, and support from other people. Chronological age alone cannot explain whether someone will accept or reject a particular technology.

Dining feedback made this individuality even clearer, while revealing differences across institutional settings. Respondents evaluated food and living conditions differently and selected different areas for improvement, from taste and portion size to infrastructure and service. At Hengyu Caojiadu, the publicly built and privately operated welfare institution among the three facilities, several responses identified portion size and taste as concerns. At the two Taikang communities, respondents more often identified infrastructure and service as areas for improvement. Reading these responses alongside the facilities' official descriptions clarified the value of direct feedback: institutional descriptions explain service models, while questionnaires can reveal concerns that are less visible at the organizational level.

These differences call for caution. Thirty questionnaires cannot establish that the type or price of a facility causes residents to have particular needs. However, one possible explanation is that people evaluate a facility in relation to both their immediate needs and the services the institution is designed to provide. When food quantity or taste is unsatisfactory, those basic concerns naturally take priority. In a community offering more extensive residential, medical, and recreational services, residents may judge the surrounding environment and quality of service more closely.

Even within the same facility, residents did not agree. A standardized "elderly meal" or service may be efficient, but it risks treating age as more important than the person.

Autonomy is ultimately the issue. The United Nations Principles for Older Persons state that older people should have independence, participation, care, self-fulfillment, and dignity (United Nations, 1991). Respecting autonomy does not mean refusing to offer assistance. It means distinguishing between support and control. A reminder can support someone who chooses to receive it; an unavoidable reminder can become an intrusion. Health monitoring can create security when a resident understands and accepts it, but it can reduce privacy when it is imposed. An AI meal-planning system can organize dietary information, but it should not quietly replace a resident's preferences with what an algorithm considers appropriate.

The differences among the three facilities also suggest that AI should not be introduced through a single institutional model. At a large continuing-care community such as Shen Garden, technology might help coordinate medical appointments, residential services, recreational activities, and long-term changes in care needs. At an urban community such as Jing'an Mansion, technology might connect health management, rehabilitation, and everyday services within a city-center setting. At a welfare institution such as Hengyu Caojiadu, systems that collect meal feedback, record individual care needs, or help staff respond efficiently might address more immediate concerns. These are possibilities to be tested with residents and staff, not prescriptions based on institutional type alone.

None of this means each institution should automatically receive more technology. The function of technology should respond to the needs identified by residents and staff. A complex platform would not be useful if residents mainly wanted larger portions or better-tasting meals. Similarly, a basic emergency button would not address every concern in a community where residents also expected support for recreation, healthcare, communication, and independent living. Technology becomes meaningful only when it responds to a clearly identified problem.

The questionnaires suggest three principles for respectful AI-assisted care. First, technology should be optional. Residents should decide whether they want AI monitoring, reminders, or companionship. Second, it should be customizable. Font size, voice guidance, reminder frequency, dietary recommendations, and privacy settings should reflect individual needs rather than age-based assumptions. Third, it should support human care rather than automatically replace it. AI may help staff organize health information or collect dining feedback, but personal conversation and human judgment remain important, especially when a resident's wishes are complex or changing.

This investigation has several limitations. The questionnaires formed a small exploratory study rather than a representative survey, and the findings cannot be generalized to all residents of the three facilities or to older adults as a whole. Because some questionnaires were completed by staff members, the responses do not represent residents' views alone. Some survey questions also used predetermined categories that may have limited what participants could express. In addition, the survey did not directly ask residents how they understood aging or whether they felt equally treated.

The institutional comparison has similar limits. Public descriptions of the three facilities were used only to understand their service models. Official websites describe how institutions present themselves; they cannot independently verify residents' experiences or explain why questionnaire responses differed. The comparison therefore provides context rather than evidence of a causal relationship between an institution's model and its residents' priorities.

My original question was what AI could do for older adults. I now believe a better question is what older adults want technology to do, and what they want it not to do. The three facilities differ in scale, location, resources, and service model, but none of those characteristics eliminates variation among the people living there. Respectful eldercare does not begin by labeling residents as dependent users of special "elderly technology." It begins by listening to them as individuals.

The future of eldercare should be designed not only for older adults but with them. Technology becomes genuinely helpful when it responds to differences rather than hiding them, supports care without automatically replacing human relationships, and gives older adults more ways to remain the authors of their own lives.