1. The 2018 Prediction Revisited
In 2018, McKnight’s published a bold forecast. It argued that artificial intelligence would transform healthcare not from within medicine’s established institutions, but from the outside. Digital forces, the piece warned, would “radically alter the way in which we operate the system.”
The response was modest. At the time, the field faced more immediate concerns: workforce shortages, reimbursement pressures, and regulatory complexity. AI seemed distant — something happening in frontier research labs, not nursing homes.
Seven years later, that prediction deserves a fresh look. Moreover, it deserves examination through the lens of long-term care — the sector that understood the coming shift long before the broader healthcare world took notice.
2. The Overlooked Sector’s Hidden Advantage
Why Long-Term Care Gets Ignored
Long-term care has always operated at the margins of healthcare’s innovation narrative. It serves the oldest and most medically complex patients. Facilities run on thin margins. They depend heavily on Medicaid reimbursement and a workforce that earns wages rarely reflecting the skill and compassion the work demands.
When venture capital imagines healthcare’s future, it pictures hospitals, precision medicine, and employer insurance markets. The nursing home down the road rarely enters that vision. Yet practitioners in this sector have long held something the innovation narrative consistently undervalues: an intimate, operational understanding of what care actually requires.
Resident-Level Knowledge Matters
This understanding develops not in the abstract, but day after day — resident by resident, family conversation by family conversation. That ground-level knowledge turns out to be exactly what responsible AI deployment in healthcare needs most. Furthermore, it is knowledge that no algorithm derived from aggregate data can fully replicate.
3. What the Forecast Got Right
External Forces Drove the Disruption
The core thesis of 2018 proved accurate on two counts. First, healthcare’s transformation came from external digital forces rather than internal reform. Second, the sector’s hierarchical, linear model of care delivery proved fundamentally incompatible with interactive, iterative digital systems.
The evidence is clear. Amazon acquired One Medical for $3.9 billion. Digital health venture capital surged from $8 billion in 2018 to $29 billion in 2021. AI start-ups reshaped clinical workflows from outside traditional institutions. Disruption arrived from exactly the direction the 2018 article predicted.
Pattern Recognition Predictions Came True
AI systems now outperform clinicians in specific radiologic tasks. They predict sepsis hours before clinical symptoms emerge. They also identify drug interactions that human review misses. The observation that digital technologies “feed on themselves” — each iteration improving on the last — anticipated the foundation models and large language systems that arrived in 2022 and changed the entire landscape of what AI could do.
4. What Needed Refinement
Large Language Models Changed Everything
Honest retrospection requires acknowledging what the 2018 analysis did not fully anticipate. The emergence of large language models shattered assumptions about task-specific AI. GPT-4 and its successors introduced general-purpose reasoning systems. These systems could converse, synthesize, and advise across domains — a capability that fundamentally altered deployment possibilities and ethical questions alike.
Data Quality Remained a Constraint
The 2018 analysis also underweighted data quality as a limiting factor. Interoperability failures slowed adoption significantly. The cost of data cleaning added friction. Critically, the absence of adequate data on underserved populations — including many people long-term care serves — introduced new disparities rather than reducing existing ones.
This challenge is one the long-term care sector is particularly positioned to understand and help solve, given how thoroughly it knows its patient populations.
COVID-19 Accelerated and Clarified
The pandemic accelerated AI adoption in ways no one predicted. Simultaneously, it demonstrated something long-term care practitioners have always known: technology augments care but cannot replace the irreplaceable human judgment and relational knowledge that direct care workers bring. COVID-19 made that truth impossible to ignore.
5. The Practitioner Advantage in AI Deployment
When AI Works Well
Long-term care’s experience with early AI deployment reveals a consistent pattern. In settings where experienced staff participate as implementation partners, outcomes improve. Their workflow knowledge, clinical judgment, and awareness of individual residents shapes how AI tools are designed and deployed.
Monitoring systems surface information that aides and nurses act on. Documentation tools free time for direct care. Predictive analytics prompt earlier interventions and prevent hospitalizations. These results appear consistently when practitioners are included in the process from the start.
When AI Falls Short
In contrast, when technology is imposed as an efficiency mechanism without genuine practitioner input, resistance emerges. This resistance is not technophobia. Instead, it reflects legitimate professional judgment. The tools fail to account for clinical complexity. They generate alerts that do not match what caregivers observe. They create new administrative burdens while claiming to reduce them.
The Wisdom Practitioner Concept
Experienced practitioners in this sector are best described as “wisdom practitioners.” Their decades of pattern recognition, contextual judgment, and relational knowledge represent exactly the intelligence that AI systems most need alongside them. Consequently, the question for this field is not whether long-term care can adapt to AI, but whether AI will be designed to enhance what practitioners bring — rather than bypass it.
6. What Comes Next for Long-Term Care
AI Is Already Here
The tipping point has passed. AI is not coming to long-term care — it is already here. Documentation platforms, monitoring systems, clinical decision support tools, and medication management systems are active in facilities today. Therefore, the real question is what kind of deployment serves residents and supports staff.
Three Key Observations
First, AI tools designed with direct care worker input consistently outperform those designed without it. This is an empirical finding, not a political preference. The people closest to residents know things about care delivery that no algorithm can fully replicate.
Second, the distinction between augmented intelligence and artificial intelligence matters enormously in long-term care. Tools that support clinical judgment — surfacing information, flagging patterns, and reducing documentation burden — have a fundamentally different relationship to care quality than tools that replace or override human judgment. In a setting defined by complexity, vulnerability, and relational trust, that distinction is not academic. It determines whether technology truly serves residents or only appears to do so.
Third, governance frameworks for AI in healthcare are still being written. This is an opportunity. Long-term care practitioners, advocates, and organizations bring the operational knowledge, ethical clarity, and track record needed to be central voices in those conversations — not afterthoughts to a process driven elsewhere.
Seven years ago, this sector saw something coming that much of healthcare had not yet faced. That same practitioner perspective — grounded, experienced, and clear-eyed — is exactly what AI deployment needs most as the field moves forward.
