The rapid integration of artificial intelligence into healthcare administration is hitting a surprising speed bump. While the initial promise of AI in the medical sector centered on slashing the bureaucratic mountain of paperwork—specifically in prior authorizations and medical coding—a new report from the Peterson Health Technology Institute paints a more complicated picture. Rather than reducing the bottom-line expenses, AI systems are in some cases inflating transaction volumes and driving up costs for health systems.

For students looking at the intersection of technology and society, this is a classic case of "Goodhart’s Law" in action: when a measure becomes a target, it ceases to be a good measure. By automating administrative tasks, health systems have inadvertently created a feedback loop. Systems are engaging in "bot wars," where automated prior authorization requests are met with automated denials, necessitating further intervention and ultimately consuming the savings that the technology was intended to generate in the first place.

The issue extends beyond simple authorization workflows into the realm of medical billing. The report highlights that the deployment of AI scribes—systems designed to listen to doctor-patient interactions and generate clinical notes—is inadvertently driving up the intensity of medical billing. These tools are often tuned to favor higher-complexity diagnostic codes, which increases costs for insurers and, by extension, patients. This phenomenon essentially automates the inflation of medical spending, creating a cycle that health plans are struggling to absorb.

Perhaps the most sobering takeaway is that this is not a failure of the software itself, but a failure of the surrounding policy and infrastructure. The current patchwork of electronic health record (EHR) systems often prevents seamless integration, leading to reliance on third-party "band-aid" solutions that add complexity rather than removing it. Researchers argue that we are attempting to bolt 21st-century intelligence onto a 20th-century administrative chassis. Without a total redesign of how these systems communicate and how incentives are structured, simply adding more AI will likely exacerbate, rather than fix, the underlying waste.

The rapid integration of artificial intelligence into healthcare administration is hitting a surprising speed bump. While the initial promise of AI in the medical sector centered on slashing the bureaucratic mountain of paperwork—specifically in prior authorizations and medical coding—a new report from the Peterson Health Technology Institute paints a more complicated picture. Rather than reducing the bottom-line expenses, AI systems are in some cases inflating transaction volumes and driving up costs for health systems.

For students looking at the intersection of technology and society, this is a classic case of "Goodhart’s Law" in action: when a measure becomes a target, it ceases to be a good measure. By automating administrative tasks, health systems have inadvertently created a feedback loop. Systems are engaging in "bot wars," where automated prior authorization requests are met with automated denials, necessitating further intervention and ultimately consuming the savings that the technology was intended to generate in the first place.

The issue extends beyond simple authorization workflows into the realm of medical billing. The report highlights that the deployment of AI scribes—systems designed to listen to doctor-patient interactions and generate clinical notes—is inadvertently driving up the intensity of medical billing. These tools are often tuned to favor higher-complexity diagnostic codes, which increases costs for insurers and, by extension, patients. This phenomenon essentially automates the inflation of medical spending, creating a cycle that health plans are struggling to absorb.

Perhaps the most sobering takeaway is that this is not a failure of the software itself, but a failure of the surrounding policy and infrastructure. The current patchwork of electronic health record (EHR) systems often prevents seamless integration, leading to reliance on third-party "band-aid" solutions that add complexity rather than removing it. Researchers argue that we are attempting to bolt 21st-century intelligence onto a 20th-century administrative chassis. Without a total redesign of how these systems communicate and how incentives are structured, simply adding more AI will likely exacerbate, rather than fix, the underlying waste.