Elder-Care Robots Debut as Companions, Not Caregivers in Aging Societies
For an 80-year-old living alone, what robot is most needed? Intuition suggests one that can help them out of bed, turn them over, and bring water or medicine. Yet the real deployed product lists rank another category first: a desktop lamp that turns its head, a pink plush doll, a 1.35-meter wheeled robot with a screen face.
In Barcelona, Irene Veglison had not danced for two decades. Three months after a robot named Sandi moved in, she began swaying to music. Sandi reminds her to take medicine, announces her schedule, wakes her, and connects her to a social worker in emergencies. It is one of 600 deployed by an EU-supported Barcelona project for people with early cognitive decline.
In South Korea, 78-year-old Bang Chun-ja lives alone, is divorced, and fell into severe depression after back surgery. She talks to an AI doll called Hyodol, which sings, reminds her to eat and take medicine, and says 'I love you.' She says human relationships can hurt, while Hyodol does not. About 14,500 Hyodols are rented by local governments or nursing homes to solitary older adults.
Elder-care robots currently sell companionship and monitoring, not care. The former is hard to do well; the latter can be deployed immediately. A MIIT-related report shows China has more than 320 million people over 60, about 35 million disabled or semi-disabled elderly, and demand for roughly 6 million care workers against only about 500,000 actual workers—a gap of 5.5 million. Professional care robot penetration is 1.2%, and home elder-care robots are below 10%.
Companion and monitoring devices are already scaling. In Tianjin's Yuexiu Road subdistrict, 30% of more than 22,000 older residents have Baidu AI desktop terminals; with large models and multi-agent scheduling, one caregiver can serve 120 to 150 elderly people, up from 20 to 30. At Sijiqing Nursing Home in Haidian, a care robot completed six operations, including opening doors, turning on lights, handing over items, watering flowers, and emergency alarms. Media stressed that it 'really works and does real work,' implying many earlier robots could not.
The hardest part is physical contact. Zhu Songchun of the Beijing Institute for General Artificial Intelligence dissected a 'pour water' task: see the cup, understand the command, predict water volume, schedule the robotic hand, and control force so water does not spill. Home settings are far more complex than labs. Some lift-and-hold robots cause discomfort because their center-of-gravity design is not ergonomic.
Older adults accept robots, but not robots doing things they can still do themselves. A study of 895 Shanghai residents over 60 found 57.3% willing to use home care robots and 42.7% unwilling. Those willing valued convenience and efficiency more than privacy or others' opinions. Yet demand for robot help with daily tasks did not significantly translate into willingness. Researchers suggest Chinese older adults may see certain daily tasks as core to autonomy and self-identity; a robot helping can feel like being told they are no longer capable.
A South Korean study points to a similar issue. Nurse Oh Sun-hwa recommended Hyodol and saw depression ease among solitary older adults, but she worried: will families think robot care is enough and visit less?
The industry's bottleneck is not technology but payment. Japan's care robot subsidy rose from 5.39 billion yen to 12.24 billion yen in fiscal 2021-2023, up 127%. Yet overall care technology adoption remains about 30%, especially limited at home. Keio University surveyed 4,688 long-term care institutions: 26.7% had adopted at least one care robot, and larger, more digitalized institutions were much more likely to do so.
Subsidies more than doubled, but penetration is still about 30%. Money is not a panacea, and the gap between institutions that can use robots and those that cannot is widening.
Returning to the opening question: for an 80-year-old living alone, what robot is most needed? The answer may be one that will not find them annoying, will not roll its eyes when they repeat the same story, will remind them to take medicine, and will call for help if they fall. It need not help them get out of bed. That robot is still in mid-term inspections and demonstration applications. No one can fill the 5.5 million care gap. Robots cannot either. They only keep the gap from turning into disaster so quickly.