Ambient AI has quickly become one of healthcare’s most compelling AI use cases.
Instead of typing notes during or after an appointment, clinicians can use ambient AI tools to listen to patient conversations and automatically generate clinical documentation. The promise is straightforward: less time staring at a screen, less administrative work, and more time focused on patients.
And there is growing evidence that it works.
A 2026 study of more than 10,000 emergency department encounters found that ambient AI use was associated with about 73 seconds less documentation time per encounter, potentially saving roughly 24 minutes during an eight-hour shift when used consistently. Other research has linked ambient clinical intelligence with reductions in after-hours documentation and clinician burnout. (Source)
But a new question is emerging: What happens when AI does such a good job documenting care that it also changes how that care gets billed?
Better Documentation, More Complexity
A September 2026 analysis from the Blue Cross Blue Shield Association found a sharp increase in patients being documented with complex conditions as hospitals adopted AI-assisted coding and documentation tools.
BCBSA estimates that increased coding complexity added approximately $942 million in healthcare spending between 2023 and 2025 for its member companies. The analysis found more secondary diagnoses being recorded, sometimes based on information such as individual laboratory values, without evidence of a corresponding increase in the care patients received. (Source)
That creates a complicated dynamic.
AI may not be inventing conditions. Instead, it can identify details already present in a patient’s record or conversation that might previously have gone undocumented. More complete documentation can improve the accuracy of a medical record.
But because documentation also influences reimbursement, capturing more clinical complexity can move an encounter into a higher-paying billing category.
In other words: better documentation can also mean a bigger bill.
The Debate Is Just Beginning
Hospitals and health systems have pushed back against the idea that AI is simply inflating bills. The American Hospital Association argues that changes in coding intensity have multiple causes, including increasing patient acuity, and that providers have legal and ethical obligations to accurately document the conditions they treat. (Source)
That distinction matters.
The emerging issue is not necessarily whether ambient AI is “good” or “bad” for healthcare costs. It is whether healthcare organizations have the governance systems needed to understand the downstream effects of these tools.
As ambient AI moves from pilots into everyday clinical workflows, health systems may need to measure more than hours saved or clinician satisfaction. They will also need to monitor changes in coding patterns, reimbursement, documentation accuracy, patient costs, and clinical outcomes.
Researchers are already examining that connection. A Johns Hopkins-led project funded in 2026 is specifically studying whether ambient AI adoption affects billing intensity and quality outcomes using physician surveys and commercial claims data. (Source)
The Next Phase of Ambient AI
Ambient AI still has the potential to solve a very real problem. Documentation burden takes clinicians away from patients and contributes to burnout.
But healthcare technology rarely changes just one part of the system.
A tool designed to save a physician time can influence documentation. Documentation can influence coding. Coding can influence reimbursement. And reimbursement ultimately affects payers, employers, and patients.
The next phase of ambient AI will therefore be about more than proving the technology works.
It will be about understanding everything that changes when it does.
