45

This article contributes to the study of sociotechnical promises and expectations by examining the rise of artificial intelligence (AI) in radiology. Drawing on a cross-analysis of two empirical datasets—a systematic examination of the professional website Thema Radiologie and a series of interviews conducted with two key stakeholder groups (radiologists and software providers)—this study identifies five major promises associated with AI adoption in radiology. These include its potential to enhance diagnostic accuracy, reduce time constraints, provide clinical assistance, improve the performance of radiological equipment, and foster the emergence of new value-generating ecosystems. The findings reveal a highly interconnected “field of promises,” characterized by a progressive shift toward greater specificity and a tendency toward “professionalization,” wherein AI-related promises increasingly align with the practical concerns of radiologists, who, in turn, reinterpret and articulate them. This evolution highlights the need for both radiologists and technology providers to engage in a critical and reflective reassessment of these promises. Such a process requires software providers to make strategic adjustments and radiologists to reject certain claims, anticipate potential limitations, and formulate alternative expectations. This study thus contributes to a broader understanding of how technological expectations are shaped, negotiated, and institutionalized within professional communities.