Log in to star items.
Accepted Paper
Paper short abstract
AI in healthcare is widely promised to transform clinical work, yet systems often break. Based on an ethnography of speech recognition, we analyze how AI systems are kept in a state where imaginaries of a future “under repair” stabilize a present “beyond repair,” leaving staff to absorb errors.
Paper long abstract
The promise of data and automation in healthcare is unbroken (Watson & Wozniak‐O’Connor, 2025). At the same time, we know that, in practice, automated, data-intensive systems constantly break, requiring extra care that most often goes unnoticed (Hertzum et al., 2025; Hoeyer, 2023). This paper examines the dynamic interplay between future-oriented promises about AI and the precarious maintenance of AI systems in the present, focusing on automatic speech recognition (ASR) in Danish hospitals. Based on an ethnographic study of both IT managers and frontline workers, we analyze a peculiar state in which, on the managerial level, current issues with ASR are neglected while deferring their ‘solution’ to future upgrades of those systems. At the same time, this leaves healthcare staff to absorb system shortcomings as clinicians internalize errors and secretaries become data validators.
We theorize this state as “meantime maintenance”. In doing so, we bridge scholarship on data work and maintenance (Bossen et al., 2019; Denis & Pontille, 2025) with the sociology of expectations (Borup et al., 2006) to analyze how digital systems are kept in a state where imaginaries of a future “under repair” stabilize a present “beyond repair.” While “meantime” has been described as a space of indeterminate possibility (Masquelier & Durham, 2023), our findings suggest that in contemporary healthcare data infrastructures, it is tightly structured by future-oriented regimes. Prospective efficiency narratives function as governance devices: expectations of imminent technological improvement stabilize present shortcomings, defer accountability, and redirect organizational attention away from maintenance needs in the here and now.
Caring for the possible: In the meantime of healthcare’s data-driven futures
Session 1