The United States currently has a shortage of roughly 100,000 nursing home workers. This shortage exists not because the work is technically complex or because the skills are rare. It exists because the work is paid too little to attract sufficient workers, is physically demanding, is socially undervalued, and offers limited career advancement. The people who need care are not being cared for because a society that has built an elaborate structure for valuing cognitive output has never built a commensurate structure for valuing care.
Meanwhile, the same economy is investing hundreds of billions of dollars in systems to automate the cognitive work that currently generates the most income. The juxtaposition is striking. We are building machines to do the work the market values most, while the work the market values least — the work that most profoundly shapes human lives — goes chronically understaffed.
What Care Work Actually Is
Care work is a broad category. It includes nursing and elder care, childcare and early education, mental health support, social work, and the informal care provided within families — the parent managing a child's disability, the adult child supporting aging parents, the partner managing a household through a health crisis.
What these forms of work share is that their value lies in the quality of the human relationship through which they're delivered. A patient in a hospital doesn't just need the right medications administered at the right intervals. They need to be seen as a person, to have their fear acknowledged, to feel that someone is genuinely attending to their wellbeing. A child in early education doesn't just need to be supervised. They need the specific kind of attentive, responsive, emotionally present engagement that builds the attachment security that shapes cognitive and social development for decades.
These are not supplementary features of care work that could be stripped away while the core function is performed mechanically. They are the core function.
Where AI Falls Short
AI systems have made real progress in care-adjacent tasks: diagnostic support, medication management, monitoring for deterioration, documentation. These are genuine contributions that reduce the cognitive burden on human care workers and can improve safety in care settings.
But the research on AI in care contexts consistently finds the same limitation: AI can monitor, flag, and recommend. It cannot provide the relational presence that makes care meaningful to the person receiving it.
A 2023 study published in the Journal of Medical Internet Research found that elderly patients in care facilities with AI-augmented monitoring reported higher objective safety outcomes but lower subjective wellbeing scores than comparable patients in facilities with higher human staffing ratios. The safety was measurable. The loneliness was also measurable, and it ran in the opposite direction.
The relational deficit of AI care is not a technical problem awaiting a better model. It is a structural feature of what AI is. A care robot that monitors vital signs and detects falls does not share mortality with the person it cares for. It does not bring its own experience of vulnerability, loss, and recovery to the bedside. The presence that matters in care is human presence — not because technology can't simulate warmth, but because the simulation isn't the thing.
The Economic Contradiction
Here is the economic contradiction at the heart of the care economy: the work that is hardest for AI to perform and that most profoundly shapes human welfare is systematically undervalued in market arrangements. The work that is most automatable commands the highest wages.
A software engineer whose work will be substantially automated within a decade earns a median salary of $130,000 in the United States. A certified nursing assistant, whose work requires genuine relational skill and whose job AI cannot meaningfully replace, earns a median of $35,000.
This inversion is not accidental. It reflects a market structure that values skills in proportion to their scarcity in formal credential systems and their legibility to employers — not in proportion to their social value or their resistance to automation. The credential system was built before anyone needed to price in AI displacement. It has not adjusted.
What Investing in Care Would Require
Revaluing care work is not simply a matter of raising wages, though wage increases are necessary. It requires restructuring the funding mechanisms for care, which are currently a patchwork of Medicaid (for the poor), private insurance (for those who can afford it), and unpaid family labor (for everyone else).
It requires recognizing informal care work — the 40 million Americans who provide unpaid care for family members — as economically productive activity worthy of support, rather than as a private family matter that occurs outside the economy.
And it requires changing the cultural narrative about what makes work valuable. The knowledge economy valorized cognitive output and treated relational labor as soft, secondary, and properly female. The AI transition is forcing a reassessment: if machines can do the cognitive output, and cannot do the relational labor, the hierarchy of value needs to be renegotiated.
This renegotiation is not merely economic. It's political, cultural, and deeply gendered — because care work has historically been feminized and devalued for reasons that long predate AI.
The Opportunity in the Disruption
There is a version of the AI transition in which the displacement of automatable cognitive work creates the space and the economic pressure to invest more seriously in the work AI cannot do. If AI captures a significant portion of the productivity gains currently embedded in knowledge work, and if those gains can be distributed broadly enough, the society that emerges might value care more, support it better, and build institutional structures — public care systems, caregiver support, paid family leave — that recognize what the work actually contributes.
That version of the transition is not automatic. It requires policy choices that are not currently being made. But the possibility exists, and understanding it requires taking the care economy seriously not as the economy's soft underbelly but as the domain where human work will remain both irreplaceable and, if society chooses, better valued than it has ever been.