Rehumanizing global health care with agentic AI
The World Health Organization projects a global shortfall of 11 million healthcare workers by 2030. This isn't a distant, theoretical problem; it's the gaping hole in the floorboards of a system already buckling under the weight of aging populations and chronic staff burnout. The industry's answer, increasingly, is to deploy "agentic AI"—not as a simple chatbot, but as an autonomous worker in its own right. Over two-thirds of healthcare providers are already using it. The pitch is compelling: le
Analysis
The World Health Organization’s prediction of an 11-million-worker shortfall by 2030 isn’t a statistic. It’s a countdown. And the healthcare industry’s desperate scramble for a solution has landed, with almost predictable inevitability, on the latest Silicon Valley darling: agentic AI. The KPMG stat is telling—68% of providers have already adopted these autonomous digital agents. It’s not a trial; it’s an invasion, marketed as a salvation. But let’s be brutally honest: we are not adopting this technology because it’s a proven, superior paradigm. We’re adopting it because we’ve failed, utterly and systemically, to value and retain the human beings who actually constitute a healthcare system.
For two decades, digitalization in healthcare has been a monument to misplaced faith in technology as a substitute for good process. Electronic health records (EHRs) were sold as the backbone of modern care. Instead, they became a digital leash, chaining clinicians to data entry and multiplying their cognitive load rather than reducing it. Telehealth and remote monitors expanded geographic access—a genuine win—but they could not replicate the nuanced, empathetic core of medicine. They solved a logistics problem while ignoring the crisis of quality and trust. Ashis Barad of HSS is right to point out these failures. But his argument that agentic AI is fundamentally different feels like a new coat of paint on a familiar, flawed house. The core promise—autonomy, iteration, supercharged workflow—is seductive precisely because it targets our most painful bottleneck: time.
The HSS case study is compelling on its face. Collapsing a multi-week, contractor-dependent claims process into an in-house AI that handles 1,100 cases a month, slashing appeal time and boosting success rates to 100%? That’s not just an efficiency gain; it’s a financial and operational transformation. It proves the technology’s potent capability to navigate complex, rule-based back-office mazes. But let’s not conflate automating insurance bureaucracy with augmenting clinical judgment. One is a problem of data processing and persistence; the other is the realm of human uncertainty, fear, and biological unpredictability. The real test, and the real danger, begins when these agents move from the back office to the bedside, as HSS is now attempting with its AI sch— presumably a scheduler or intake tool.
Here’s my sharp dissent: Agentic AI is being rushed into a vacuum of trust and systemic failure. We are using it as a pressure valve for a system hemorrhaging staff due to burnout, poor compensation, and a loss of autonomy. By deploying agents to triage patients or manage non-clinical interactions, we risk making a fatal trade. We are automating the interfaces that build trust and gather context—the very "inefficiencies" of a human conversation that allow a nurse to hear anxiety in a patient’s voice or a clerk to notice confusion about medication. To digitize that is to risk further alienating patients from a system that already feels cold and fragmented.
Furthermore, the "autonomy" of these agents is a double-edged sword. In a domain governed by strict privacy laws (like HIPAA) and life-or-death decisions, autonomous decision-making is a terrifying prospect. Who is liable when an AI agent misinterprets a symptom during triage and advises incorrectly? The vendor? The hospital that deployed it? The clinician who signed off on the protocol? The legal and ethical frameworks are utterly inadequate for this new reality. We are building a dependency on systems whose decision-making processes are often opaque "black boxes," creating a new class of risk that hospitals are ill-equipped to manage.
The most provocative thought is this: perhaps agentic AI is the ultimate monument to our unwillingness to address the root cause. Instead of fighting for better pay, mental health support, and a reduction in the administrative parasitism that drains clinicians, we are building a digital workforce to do the jobs humans no longer want. It’s a tech-solutionist answer to a human-resource problem. Yes, the technology is impressive. Yes, it will likely save money on claims processing. But as we celebrate collapsing workflows, we must ask: are we collapsing the very human element that makes healthcare more than a transaction? We are building a system that needs fewer humans by design, not just by circumstance. That isn't innovation; it's a capitulation. The 11-million-worker gap will not be filled by software agents, no matter how sophisticated. It will be filled by a recommitment to the human beings who chose this profession to care for others—a choice we are currently rewarding with burnout and a digital replacement.
Disclaimer: The above content is generated by AI and is for reference only.