Rethinking "Useful" and "Useless" AI in radiology

Artificial Intelligence (AI) is reaching a pivotal moment in radiology, with rapidly expanding applications that often overwhelm clinicians and fuel skepticism or fear of missing out. This editorial proposes a simplified, practice-oriented framework that categorizes AI tools as either useful or useless based on their true added value to radiologists. Useless AI comprises tools that replicate tasks radiologists already perform efficiently, offering little clinical benefit. In contrast, useful AI addresses tasks beyond human capability, such as radiomics, large-scale text processing, and highvolume image analysis. By clarifying this distinction, we encourage radiologists to adopt AI judiciously, defend core clinical skills, and prepare for future technological advances.