Artificial intelligence tools are increasingly deployed in radiology departments worldwide, yet adoption remains uneven. A prevailing assumption attributes this gap to radiologists' psychological resistance — a so-called "fear of AI." This study challenges that narrative directly.

Using the Technology Acceptance Model (TAM), we surveyed all radiologists in a high-volume academic department and combined their responses with real-world operational data from 46,000 AI-assisted examinations. The goal was to identify what actually drives — or blocks — sustained AI integration in a mature clinical setting.

The barrier is not fear, seniority, or professional identity. It is software quality and system reliability.