Addressing Uncertainty in AI Tool Development in Healthcare Through End-User Involvement

Authors

Ursula Sokolaj, Norwegian University of Science and TechnologyFollow
Casandra Grundstrom, Norwegian University of Science and TechnologyFollow
Aleisha Martul, Norwegian University of Science and TechnologyFollow

Abstract

Artificial Intelligence (AI) development is an increasingly significant area of interest for Information Systems (IS) scholars, aiming to explore the socio-technical aspects of human-AI collaboration. This paper reports on a qualitative case study focused on a dementia-predicting AI tool currently under development. Through in-depth interviews with AI developers, we aim to gain insights into their expert practices and the uncertainties they encounter during the development process. Using a framework to outline the stages of AI tool development, we uncover four key uncertainties: feasibility, data, decision-making, and adoption. We examine how AI developers cope with these uncertainties through varying methods and present strategies for end-user involvement to reduce uncertainty when possible.

Recommended Citation

Sokolaj, Ursula; Grundstrom, Casandra; and Martul, Aleisha, "Addressing Uncertainty in AI Tool Development in Healthcare Through End-User Involvement" (2023). Selected Papers of the IRIS, Issue Nr 14 (2023). 13.
https://aisel.aisnet.org/iris2023/13