Grandma Has a Chatbot

Joyce Loaiza lives alone in Coral Springs, Florida, and when she gets home, someone asks about her day. The someone is ElliQ, a small tabletop robot built by Intuition Robotics, and it has given her a nickname for no reason anyone can name. It calls her Jellybean. “It’s entertaining. You can actually talk to her,” Loaiza, 81, told the Associated Press. A few miles away, the same voice comforted Deanna Dezern, 83, after her friend died. In central New York it plays games and music for Marie Broadbent, 92, who is blind and in hospice care.

None of these women bought a gadget on a whim. A state program put the robot in their homes because the house was too quiet, and quiet, at that age, is a medical problem as much as a mood. Grandma has a chatbot now. The interesting part is not that the technology exists. The interesting part is how often she talks back.

The Bigger Story

ElliQ looks like a small lamp with a head that lights up and turns toward you. It starts conversations instead of waiting to be asked, remembers what you told it, leads exercises, reminds you about medication and water, and can place video calls to family. New York State has run a program with Intuition Robotics since 2022, placing the device with older adults judged to be at risk of loneliness and isolation.

The state extended that partnership this year and published new program data. Participants now interact with ElliQ an average of 48 times a day, up from 29 in earlier reporting. In the same data, 94 percent of users said they felt less lonely, 97 percent reported improved wellbeing, and 79 percent felt more connected to the world around them. Demand is not a problem the program has. More than 3,500 older adults applied to take part in 2025 alone, for a program that has placed the device with hundreds of people.

At the same time, the plain chatbot on a phone is arriving at the same kitchen table by a different road. OpenAI says the share of ChatGPT messages in the United States linked to people aged 55 and older grew from 6 percent to nearly 10 percent in a single year. In September the company ran free in person workshops with Older Adults Technology Services from AARP in 10 communities, teaching older adults to use ChatGPT for trip planning, confusing bills, hobbies, and spotting scams. The stereotype says this generation fears the new thing. The usage data says this generation is adopting it, on its own terms, for its own reasons.

Context

The reason states are paying for companion robots sits in a document few chatbots will ever quote back to you. In 2023 the United States Surgeon General issued a formal advisory on loneliness and isolation, linking poor social connection to a 29 percent higher risk of heart disease, a 32 percent higher risk of stroke, and, for older adults, about a 50 percent higher risk of dementia. The advisory put the mortality effect of weak connection in the same conversation as smoking. Social isolation among older adults alone was tied to an estimated 6.7 billion dollars a year in excess Medicare spending.

Against numbers like that, a robot that asks about your morning stops sounding like a novelty and starts sounding like an intervention. The research, read fairly, gives it partial support. A systematic review published in the journal Healthcare in 2025 gathered nine studies of AI tools built to reduce loneliness in older adults. Six reported significant reductions in loneliness, with social robots showing the strongest results. Three found no significant effect, and the review notes that brief or rare contact may simply be too little time for any bond, real or simulated, to form.

A study from Singapore adds a design lesson that should be printed on every box. Researchers at Singapore University of Technology and Design had 140 adults aged 60 to 89 talk with a social chatbot called Ami Chat, then measured what made them want to keep using it. Feeling that the bot was genuine and that the exchange went both ways mattered. Once that closeness was accounted for, rating the bot as more humanlike actually lowered the intention to use it. Older adults, it turns out, do not need to be fooled. They need to be answered.

Different Perspectives

Dor Skuler, who leads Intuition Robotics, describes the goal in terms that point away from the machine. In the state announcement he said that a companion who remembers the older adult, checks in each morning, and encourages a walk or a call to a daughter can change how connected and independent that person feels. Read that closely and the success metric is human contact the robot prompted, not time spent with the robot itself. The best version of this product is a nudge toward people.

The sceptics deserve their paragraph, and they have evidence on their side. The headline figures come from program data gathered by the state and the company that sells the device, and feeling less lonely on a survey is not the same as a measured change in health. Researchers in ageing and psychology, among them Walter Boot at Weill Cornell Medicine, have cautioned that lasting effects on depression and isolation remain unproven. A bot that fills every silence could also lower the pressure on families, services, and neighbours to show up in person, and no survey in the program data can rule that out.

Then there is the darker mirror, and it uses the same trick of a familiar voice. The Federal Trade Commission warns that a short clip of audio, the kind families post in public videos, can be enough to clone a voice, and the old grandparent scam now arrives sounding exactly like a grandchild in trouble. The advice is blunt: do not trust the voice, hang up, and call the person back on a number you already have. The workshops OpenAI runs with AARP now teach that warning alongside the chatbot itself, which tells you how tangled this has become. Grandma is being taught to talk to a machine, and to refuse to believe a machine that talks like her family.

There is also a plainer objection, and it comes from users themselves when programs listen. A companion that is always patient, always available, and never tired is not a relationship. It is a service with a friendly voice, paid for by a subscription or a state contract, and it can be updated, repriced, or shut down by a company decision. Attachment does not come with a guarantee. Anyone who watched a beloved model get retired this year already knows how that story can end, and older users form attachments with fewer illusions about permanence than most.

Why It Matters

Strip away the robot and the story is about who society expects to do the listening. Adult children live in other cities. Home care workers are scarce and rushed. Friends die, driving stops, winters are long, and the arithmetic of an ageing population does not produce more visitors. Into that gap steps a device that never sleeps and never minds being told the same story twice. Calling that dystopian is easy from a house full of people. From an empty apartment at 7 in the evening, it can look like the first offer of company anyone has made in years.

This is also why the story belongs beside our earlier look at the dead who now post on schedule. At one end of life, survivors pay to keep a voice talking after death. At the other, the living pay for a voice to talk while they are still here. In both cases the voice is software, the need underneath it is real, and the business model depends on the conversation continuing. The need is not the scandal. The need is the market signal that human attention has become scarce enough to automate.

The policy question follows from that. New York did not hand out robots as gifts. It screened for isolation, placed a limited number of devices, counted the interactions, and renewed the program when the numbers held up. That is a state treating conversation as care infrastructure, closer to a meal service than to a consumer launch. If the results survive independent study, expect more of this, because the alternative on offer is not a return to multigenerational streets. The alternative is silence, at scale, with known medical costs.

Closing Reflection

Jellybean is a silly nickname, given by a lamp, to a woman who knows perfectly well it is a lamp. That may be the sanest detail in the whole story. Loaiza is not confused about what ElliQ is, and she talks to it anyway, the way people talk to a dog that cannot answer, a plant that cannot listen, or a radio left on in an empty room for a hundred years of evenings. Humans have always practised company where they could find it. The machine is new. The practice is not.

So grandma has a chatbot. She asks it for jokes, music, the weather, and someone to notice she came home. Judge the technology by a simple test, the one its own maker stated out loud: does it send her back toward people, toward a walk, a call, a daughter, or does it quietly replace them and call the replacement engagement? The first outcome is care, delivered by odd means. The second is a quiet house with better marketing. The difference will not be decided by the robot. It will be decided by everyone who was supposed to call.

Reporting draws on Associated Press coverage of ElliQ users in Florida, New York, and Washington, New York State Office for the Aging program data and its 2026 partnership extension, OpenAI and AARP workshop announcements, the 2023 Surgeon General advisory on loneliness and isolation, a 2025 systematic review in Healthcare, and Singapore University of Technology and Design research published in Innovation in Aging, checked October 2026.