On the exhibition floor, vendors promoted technologies that could monitor residents, identify potential problems, alert caregivers, and reduce staff workload. The possibilities were impressive, particularly for an industry struggling with workforce shortages and rising costs.

At the keynote presentation, Ethan Mollick, a Wharton professor and author of two books on AI, described a technological revolution that extends far beyond monitoring and automation. AI systems can now conduct research, develop plans, create software, and undertake substantial projects with surprisingly little human intervention. Their capabilities are expanding at a remarkable rate.

I've been working extensively with AI myself, so I shared much of his enthusiasm. But Mollick also emphasized something that could easily be lost amid the excitement: organizations must deliberately preserve human judgment, meaningful relationships, and accountability as they adopt these technologies. I left wondering whether the executives most drawn to AI's promised efficiencies fully appreciated that warning.

The importance of personal experience

Executives are accustomed to delegating technical matters. They rely on specialists, evaluate proposals, approve budgets, and hold managers responsible for results. That approach generally works, but AI presents an unusual challenge. Generative AI can perform an extraordinarily difficult task successfully and then make a surprising mistake on something elementary. It can produce a polished, persuasive answer that contains subtle errors or omits important considerations. Increasingly autonomous agents can take on complex assignments without human supervisors watching every step.

Mollick calls this uneven distribution of abilities the jagged frontier. The difficulty is that success on one assignment doesn't reliably predict success on another. I've encountered that frontier repeatedly in my own work. AI has helped me research complicated issues, develop essays, analyze organizational policies, and build technical solutions. It has saved me countless hours and made projects possible that I might never have attempted alone. Yet I have also learned to question its assumptions, check its conclusions, and occasionally discard an approach and begin again.

What makes this productive is the collaboration. I bring knowledge, experience, objectives, and judgment to the work. AI brings extraordinary capabilities for research, analysis, and production. Neither contribution eliminates the need for the other. Mollick urged executives to use AI themselves rather than leaving the learning to their technology departments. I think that's essential. Leaders don't need to become programmers, but without some firsthand experience they may have little basis for judging what the technology can reliably accomplish. An impressive demonstration is a poor substitute for understanding.

Efficiency, caring, and the right to privacy

The AI monitoring technologies exhibited at LeadingAge brought this concern into focus. I am an outside observer of these systems, not someone qualified or intending to evaluate their clinical or technical performance. I can readily imagine their benefits, particularly when early detection lets caregivers respond before a resident suffers serious harm.

What interests me is how organizations will use the time and resources these technologies might save, and what residents may be expected to surrender in exchange. If electronic monitoring relieves caregivers of routine observation and documentation, perhaps they can devote more attention to residents. That would be a welcome development. But if the principal objective becomes reducing staffing, the result may be quite different.

Several demonstrations also included an aspect I found distinctly unsettling. Some systems collected detailed information about how “Mom” or “Dad” spent the day and made it available through software that allowed family members to evaluate how their loved ones were doing. As an independent resident of a Life Plan Community, I do not want an AI surveillance system to track and report my daily activities to others. Where I go, how I spend my time, and when I choose to be alone are ordinarily my business. The fact that I am an older adult does not make me less entitled to privacy.

There was something ominously Big Brother about these demonstrations, compounded by an assumption that adult children should be able to oversee their parents' daily lives. It struck me as a form of ageism, however benevolent the intentions might be. I recognize that some residents, particularly those experiencing cognitive impairment or other conditions that compromise their safety, may benefit from closer monitoring. Families understandably want reassurance about the well-being of people they love. But the need for monitoring should be established individually, with meaningful consent wherever possible, clear limits on what is collected, and appropriate safeguards governing who receives the information. The ability to monitor someone does not, by itself, establish a right to do so.

This brings me to a distinction I believe deserves attention throughout eldercare: technology can make a community more efficient without necessarily making it a better place to live.

We should ask how AI affects the quality of care, residents' independence and dignity, and the relationships that make a community more than a collection of services. Savings in labor costs and improved monitoring tell only part of that story.

The danger of misplaced confidence

Shortly before the conference, I encountered an attractive brochure from an organization called Fully Informed Conversations. It offered personalized assistance with advance directives and end-of-life decisions through two named advisers, Fiona and Victor.

The brochure was professionally designed and reassuring. At first, I was impressed, but something about its promises made me uneasy. When I investigated further, I learned the advisers were AI agents, a fact the brochure did not prominently disclose. I also found reasons to question whether the brochure described the services with sufficient attention to their limitations.

AI can be a valuable source of information about complex subjects, including end-of-life planning. But anyone seeking personal guidance on consequential medical, legal, or ethical questions needs to understand the adviser's nature and limitations. My own initial reaction to the brochure is revealing. Despite my experience with AI, I let its polished presentation influence my judgment before examining what was actually being offered.

Something similar occurred during Mollick's keynote demonstrations. He displayed elaborate plans, reports, and spreadsheets generated by AI in remarkably little time. Much of the text was too small for the audience to read, so although we saw the speed and volume of production, we couldn’t evaluate the substance. Mollick was demonstrating capabilities, and he made his point. But the experience reminded me how easily we can mistake an impressive product for a useful one. The more convincing AI becomes, the more important it is to examine what lies beneath the presentation. That is especially true when vulnerable people may rely upon the information or services being offered.

Accountability remains a human responsibility

Mollick challenged organizational leaders to think beyond using AI to produce more reports, documents, and administrative output. Instead, they should reconsider how work is organized, identify activities that are no longer necessary, and use the technology to achieve more ambitious goals.

I find that an exciting prospect. In my volunteer work with the National Continuing Care Residents Association, AI has already made it possible to undertake research, communications, and technical projects with resources that would otherwise be inadequate. It has expanded what we can accomplish without increasing the demands on our volunteers. But efficiency must serve a purpose. Eliminating unnecessary administrative work is worthwhile because it frees people to focus on more important responsibilities.

For organizations serving older people, those responsibilities include caring relationships, informed decision-making, respect for personal autonomy, and the preservation of trust. That brings the discussion squarely into the realm of governance. Executives and trustees have obligations extending beyond approving a technology purchase and accepting assurances from its vendor. Their fiduciary responsibilities and stewardship obligations require them to understand the consequences of the systems they authorize. They should know what the technology is expected to accomplish, how its performance will be evaluated, what happens when it makes mistakes, and who remains responsible for decisions affecting residents. They must also ask whose interests a particular technology serves, what personal information it collects, who can see that information, and whether residents have a meaningful choice about participating. These are questions of leadership, not merely technical implementation. A board need not master the underlying technology to exercise responsible oversight. It must, however, insist on understandable evidence, clearly assigned accountability, and an honest assessment of both benefits and risks. That assessment should explicitly respect residents' independence and privacy.

Mollick closed his keynote by urging leaders to experiment and help shape a future in which AI strengthens human experience. I share that aspiration. My own experience has convinced me that the technology's potential is extraordinary, but realizing it requires knowledgeable people who remain actively engaged in deciding how it is used.

For eldercare organizations, I would put the challenge this way: What work can AI assume so that people have more time and greater capacity to do the work that gives caring its meaning, while preserving the independence and dignity of those they serve?

That seems a more promising ambition than simply finding ways to operate with fewer people or monitor residents more closely.

AI will undoubtedly take on more of the work we now perform. As human leaders, responsibility for the people affected by that work remains ours.

This is the first of three articles prompted by my experience at the 2026 LeadingAge Annual Meeting. Two companion articles will follow: a brief summary of Ethan Mollick's keynote, “Co-Intelligence: Leading in the Age of AI,” and a fuller review of his presentation and its implications for organizational leadership. The latter will examine the research, examples, and questions that lie behind the concerns I've raised here. To round-out the series, I will post an essay that shares my personal journey into collaboration with AI. Please comment here or in the forums.

Return to Previous Page