From AI to AID: Algorithmic imaging detection (AID) as a more precise label for AI in radiology.
Authors
Affiliations (3)
Affiliations (3)
- School of Medicine, Royal College of Surgeons in Ireland, Dublin, Ireland. Electronic address: [email protected].
- University of Virginia Hospital, Department of Internal Medicine, USA.
- Department of Radiology, Los Angeles General Hospital, Los Angeles, CA, 90033, USA.
Abstract
The term artificial intelligence (AI) encompasses a heterogeneous range of computational methodologies that differ substantially in function, degree of autonomy, and evidentiary support. In patient-facing communication, this heterogeneity and associated uncertainty may introduce clinically relevant interpretive and liability concerns. The designation "AI" alone provides limited information regarding a system's specific function, the extent of clinician oversight and verification, or the rigor and scope of its validation. Moreover, emerging evidence indicates that diagnostic terminology can causally influence patients' perceived risk, expectations regarding care, and treatment preferences. We therefore propose targeted modifications to patient-facing materials, informed-consent discussions, and clinical report annotations to improve terminological precision and transparency. For non-imaging applications, analogous function-specific descriptors could be adopted. Importantly, terminology should complement, rather than substitute for, the substantive disclosures that patients consider most relevant, including system function, clinician involvement, validation, performance characteristics, and limitations.