Artificial intelligence is often presented as a solution to clinician workload, promising to reduce documentation, support clinical decisions and return more time to patient care. A recent editorial in The American Journal of Medicine cautions that poorly implemented AI may instead introduce a new form of cognitive burden.
The authors use the emerging term “AI brain fry” to describe mental fatigue associated with repeatedly using, supervising and verifying AI-generated outputs. Rather than removing work, AI may shift clinicians into new roles requiring constant fact-checking, error detection, response integration and oversight.
This distinction is important in healthcare. Even highly capable AI systems can generate inaccurate, incomplete or contextually inappropriate recommendations. Clinicians therefore remain responsible for reviewing outputs while simultaneously managing patients, electronic medical records, alerts and existing administrative demands. The result may be less cognitive relief than anticipated.
AI has already demonstrated meaningful benefits. Ambient clinical scribes may reduce documentation burden, while AI-supported imaging and endoscopy systems can improve aspects of diagnostic performance. However, these gains should not lead health services to assume that every AI implementation will reduce workload.
The authors argue that healthcare organisations should assess AI not only by technical accuracy or time saved, but also by its effects on clinician attention, fatigue, autonomy and clinical skills. Poorly designed systems may contribute to alert fatigue, automation bias, overreliance and gradual deskilling.
The central message is that successful healthcare AI should reduce total cognitive workload—not simply replace one task with several new oversight responsibilities. Implementation should therefore include careful workflow design, clinician training, monitoring of unintended consequences and continued opportunities for unaided clinical practice.
As AI becomes embedded across healthcare, protecting the cognitive wellbeing of clinicians will be essential to ensuring that the technology improves rather than complicates patient care.
Read more: https://www.amjmed.com/article/S0002-9343(26)00383-9/fulltext