Tom is 78, a Marine veteran with early-stage dementia who lives alone in a ranch house outside Raleigh. His daughter, Lisa, set him up with an AI companion device last fall, a sleek tablet that reminds him to take his blood pressure medication, asks about his mood, and plays Glenn Miller when he seems agitated. For three months, it worked beautifully. Then one Tuesday afternoon, Tom fell in his kitchen. The device, registering no voice interaction, eventually prompted: “Tom, would you like to hear a joke?” Lisa didn’t find out until she checked in after leaving work four hours later.
Tom’s story is quickly becoming America’s story. More than 70 million baby boomers are alive today, and in 2026, the oldest are turning 80. They’re entering the most medically complex years of their lives – juggling multiple chronic conditions, cascading medications, and cognitive decline – just as the physicians trained to navigate that complexity are vanishing. The United States has roughly 7,000 board-certified geriatricians, fewer than one for every 10,000 older adults. In rural counties, the number is often zero. Geriatric fellowship programs rank among the lowest-filled in all of medicine, partly because geriatricians earn $20,000 less annually than internists who skip the extra training entirely.
Into this vacuum rushes Silicon Valley. The AI-in-aging-care market is projected to reach $322 billion by 2034, growing at over 21 percent annually. CBS News recently profiled robots rolling into nursing homes to combat staffing shortages. The pitch is irresistible, that technology will fill the gap that humans can’t.
But I think there is a big difference between filling a gap and papering over a wound. Sure, AI can assist elder care. Predictive fall detection, medication management algorithms, and remote monitoring systems are genuine innovations. But for me the question is whether we’re deploying these tools to enhance human relationships or to try to replace them.
The data suggest we’re trending toward the wrong answer. A George Mason University study published this February found that nearly half of dementia caregivers report low confidence in their ability to use digital health tools, even when they’re already using patient portals and telehealth regularly. Confidence drops further when caregivers must manage those systems on behalf of someone else, which is precisely what dementia care demands. We are building cockpits and handing the controls to people who were never offered flight school.
Meanwhile, the 63 million Americans serving as family caregivers are drowning. Seventy-eight percent report burnout. Only 11 percent have received any formal training. One in five says their own health is suffering. These are the people Silicon Valley’s pitch decks quietly assume will integrate new technologies into already overwhelming days. Lisa doesn’t need Tom’s tablet to tell a joke. She needs someone trained and trustworthy sitting with her father while she works a twelve-hour shift.
This is where I think policy has to catch up to the marketing. CMS should create reimbursement pathways that reward hybrid care models, pairing AI-driven monitoring with trained human caregivers who actually know the patient. An algorithm can flag an irregular heartbeat. But it cannot notice that Tom hasn’t mentioned his late wife in two weeks, or that he’s wearing the same shirt for the fourth day running. Those observations require a person in the room, not a sensor on the wall.
Congress should also invest directly in the geriatric workforce pipeline, expanding loan forgiveness for geriatric medicine fellows, funding intergenerational caregiver training programs that connect health professions students with aging families, and creating a federal caregiver tax credit that acknowledges the $870 billion in unpaid care family members already provide.
The demographic math is going to be unforgiving for our society very soon. The population aged 65 and older will grow steadily through 2036 and beyond, and no amount of venture capital changes the fact that aging is a fundamentally human experience. It requires presence, patience, and the kind of judgment that no large language model yet possesses – the judgment to know when a frail man on the kitchen floor doesn’t need a joke, but that he needs a hand.
We can build the most sophisticated elder care AI on the planet. But if we let it become a substitute for human connection rather than a scaffold for it, we won’t have solved the caregiving crisis, we’ll just have made loneliness more efficient.
Photo: delihayat, Getty Images
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.