Artificial intelligence is rapidly reshaping healthcare, but many of the industry’s biggest operational problems remain unresolved. Hospitals continue facing administrative bottlenecks. Providers still struggle with documentation burdens. Insurers continue navigating authorization complexity. Patients are still waiting weeks or months for treatment approvals that directly affect their recovery and quality of life.
The scale of the challenge is enormous. Healthcare systems globally are facing mounting workforce shortages and rapidly expanding digital demands that continue to reshape how care is delivered and managed. At the same time, healthcare systems are still operating across fragmented data environments where secure, interoperable, real-time information sharing remains a growing challenge. Just 19% of healthcare organizations expressed confidence in their ability to comply with AI-related regulations, while only 24% reported confidence in managing evolving privacy and data requirements across increasingly disconnected systems.
The financial consequences of inefficiency continue escalating. Health insurers deny roughly 850 million claims each year in the United States, creating massive administrative strain across providers, payers, and care teams. Many of those denials are later overturned after prolonged appeals and reconsideration processes, resulting in billions of dollars and countless hours spent navigating delays tied to missing documentation, fragmented systems, and disconnected decision-making processes rather than disagreement over the actual medical evidence.
Yet despite the growing investment in healthcare AI, I believe the industry is still solving the wrong layer of the problem.
Most healthcare AI systems today are being developed by highly skilled engineers and technologists. These teams understand large language models, automation frameworks, and machine learning exceptionally well. The issue is that healthcare is not simply a technical environment. It is a deeply contextual system built around clinical evidence, jurisdictional requirements, payer criteria, treatment guidelines, documentation standards, and constantly evolving decision pathways.
Without that context, even sophisticated AI systems can struggle to function effectively inside real healthcare workflows.
On the other side, many healthcare organizations attempt to build AI solutions internally because they deeply understand clinical workflows and the realities of patient care. But despite that operational expertise, many still lack the engineering specialization required to build scalable AI systems capable of functioning consistently across massive volumes of unstructured information.
What the industry has created is a disconnect between technical intelligence and domain intelligence. One side understands AI but lacks the clinical depth. The other understands healthcare but struggles to operationalize AI infrastructure effectively. Until those two capabilities are genuinely integrated, healthcare will continue producing systems that generate information without fully understanding how healthcare decisions actually get made.
The core issue is not that healthcare lacks information. In reality, healthcare may be one of the most information-heavy industries in existence.
Healthcare already operates within a massive and constantly evolving body of information that many professionals struggle to navigate efficiently in real time. In workers’ compensation alone, decision-making often depends on thousands of changing requirements spread across jurisdictions, insurers, and treatment standards, creating a level of complexity that can easily slow approvals and delay care.
The problem is finding the right information for the right decision at the exact moment it is needed. That missing context layer is where healthcare AI continues to fall short.
Too many systems focus on producing outputs without understanding the conditions surrounding the decision itself. Healthcare decisions are rarely binary. A treatment may be appropriate for one patient but inappropriate for another. Documentation requirements may differ depending on the insurer, diagnosis, injury mechanism, prior treatment history, or state-specific regulations involved.
Without understanding that context, delays become inevitable. Hospital systems and provider groups now dedicate entire departments to navigating authorization workflows, claims requirements, and medical documentation reviews. Physicians often document care inside electronic medical record systems that may not align with what insurers specifically require to authorize treatment.
When critical information is missing, treatment requests can quickly enter prolonged cycles of denials, appeals, and repeated documentation reviews, turning what should be timely decisions into delays that can stretch for months.
Most of these delays are not driven by malicious intent. Providers generally want patients to recover. Employers want injured workers to return safely. Insurers want appropriate treatment delivered efficiently and responsibly. But every stakeholder is often working from fragmented information spread across disconnected systems.
That fragmentation becomes especially dangerous in workers’ compensation environments, where injured employees are frequently unable to work while treatment decisions remain unresolved.
These are not abstract administrative delays. These are people trying to support families, maintain financial stability, and recover from injuries sustained while doing their jobs. The consequences of prolonged delays can extend far beyond physical recovery.
The human impact of these delays can become difficult to reverse over time. Research examining return-to-work outcomes after work-related injuries found that prolonged absence from employment was closely associated with poorer recovery outcomes and greater difficulty returning to sustained work participation. The same analysis also identified chronic pain, psychological distress, and reduced functional capacity among the factors that increasingly affect workers the longer recovery and treatment delays continue.
What concerns me even more is the growing knowledge gap emerging across healthcare systems.
Many of the professionals who understand utilization review processes, claims systems, clinical guidelines, and payer requirements at a deep operational level are approaching retirement. Replacing that institutional knowledge takes years. New employees entering the field face an overwhelming learning curve because healthcare decision-making involves thousands of evolving standards spread across countless systems and jurisdictions.
This is where AI can become genuinely transformative, but only if it is designed to augment human judgment rather than bypass it.
Healthcare does not need more tools that simply generate information faster. It needs systems capable of connecting evidence, clinical context, workflow requirements, and decision-making pathways in ways that help people make informed choices before delays escalate into larger problems.
Most importantly, healthcare technology should empower patients themselves.
Patients should not spend months trapped between providers, insurers, and administrative systems without understanding why treatment was delayed or what information is still required. They deserve transparency around the evidence guiding decisions about their care. They deserve access to understandable information that allows them to participate meaningfully in decisions affecting their long-term well-being.
When healthcare AI focuses only on automation, it misses the human reality at the center of every claim, authorization request, and treatment decision. But when AI helps connect the right information, the right evidence, and the right context at the right time, it can do something far more important than improving efficiency.
It can help patients reclaim agency over their own healthcare at the moment they need it most.