Pieces of an AI: Repair in Mammography Screening in Denmark
Pieces of an AI:
Repair in Mammography Screening in Denmark
Professor Dorthe Brogård Kristensen, Department of Business Management, 糖果派对
Wednesday, May 15, 14:00-16:00, Campus Odense, room U57
(online participation: https://syddanskuni.zoom.us/j/68084267660)
This presentation focuses on the implementation of AI in mammography screening as repair work. The explicit goal of repair is to reduce the workload of health professionals and there is already some evidence that this kind of repair is efficient. In a Danish report examining AI use in a breast cancer screening program over a period of three years, the combination of AI and specialised radiologists found significantly more cases of breast cancer, while reducing the number of false positives. The conclusion from the report was that AI technology is the solution to the current lack of healthcare personnel, promoting a broader crisis of healthcare systems.
From earlier studies we know, however, that the implementation of new technologies is a complex task in healthcare organizations: it might take years for practices to change. Inspired by the work of Suchman (2007, 2023) and Amoore (2023), this article seeks to explore the practical reality in the clinique where political and economic issues, such as personnel shortages, are seen as solvable through an AI system.
I will present data from a case of implementation of AI in mammography screening in four Danish hospitals. Over a period of one year, we interviewed hospital managers and radiologists in healthcare settings where the implementation is currently taking place. Based on this data, the goal is to analyse the mammography screening repair work by focusing on the arrangement of human/ machinic forms of expertise, the medical evidence concerning AI use in screening and the emerging tensions and challenges. The article pieces together different aspects of this complex reality, including 1) medical evidence and mortality 2) the effects of implementation of AI on workflow 3) notions of responsibility and accountability in the work division between the radiologist and the algorithmic system. Together, these aspects highlight what counts as expertise in mammography screening and what kinds of expertise is required. In particular, we will highlight the emerging role of the radiologist as a quality controller of an AI system.
This study makes it clear that AI solutions are never simple repair work but carry with them a number of implications that need to be considered. In particular, we pay attention to the possibility that repair efforts promote breakages that call for further repair. In this case, this means that new types of knowledge and algorithmic competences, which are not necessarily readily available, are needed. With the notion of reflexive repair, we contribute to discussions of how repair can be thought of as a productive force, without leading to unintended changes in workflow and expertise.
References:
Amoore, L. (2023). Machine learning political orders. Review of International Studies, 49(1): 20-36.
Bruun, M. & Krause-Jensen, J. (2022). Inside Technology Organisations: Imaginaries of
Digitalisation at Work. In The Palgrave Handbook of the Anthropology of Technology. Pp. 485-
505. Singapore: Springer Nature Singapore.
Burrell, J. (2016). How the machine ‘thinks’: Understanding opacity in machine learning
algorithms. Big data & society, 3(1): 1-12.
Noordegraaf, M. (2020). Protective or connective professionalism? How connected professionals
can (still) act as autonomous and authoritative experts. Journal of professions and
organization, 7(2), 205-223.
Suchman, L. (2007). Human-Machine Reconfigurations: Plans and Situated Actions (2nd ed.).
Cambridge University Press.
Suchman, L. (2023). The uncontroversial ‘thingness’ of AI. Big Data & Society, 10(2).
https://doi.org/10.1177/20539517231206794
Author info:
Dorthe Brogård Kristensen
Department of Business & Management
University of Southern Denmark
Email: dbk@sam.sdu.dk
Minna Ruckenstein
University of Helsinki
Email: minna.ruckenstein@helsinki.fi