The Effect of AI on the Radiologist Workforce: A Task-Based Analysis
Medrxiv, 2025
This preprint develops a quantitative, task-based model to estimate how AI applications are likely to affect the radiologist workforce over the next five years. The study breaks down the daily work of radiologists into specific tasks and evaluates how existing or foreseeable AI tools could influence each task category.
Using published literature and expert judgement where data were limited, the authors first identified the core tasks that occupy radiologists’ time, such as performing and interpreting studies, protocoling, communication, and reporting. They then mapped a set of AI applications (e.g., study delegation, triage, abnormality detection, automated report drafting) to those tasks to estimate task-level effects over a horizon of five years.
The model projects that AI could reduce the total hours worked by radiologists by about 33% within five years, with a plausible range of 14% to 49% reduction depending on adoption and specific tool effects. The most significant impacts come from automated report drafting across all imaging modalities and delegation of routine radiography and mammography interpretation tasks to AI.
Despite this substantial reduction in hours worked, the analysis suggests that radiologist job loss is unlikely in the foreseeable future, because workforce numbers are relatively stable, imaging volumes continue to grow, and many radiology tasks still require human expertise.