From Half-Truths to Situated Truths: Exploring Situatedness in Human-AI Collaborative Decision-Making in the Medical Context

Abstract

While the introduction of artificial intelligence (AI) solutions has large potential to improve organizational decision-making, it requires a further understanding of how humans and AI can collaborate. Through the lens of situatedness, this paper attempts to provide insight into the wider nature of human-AI collaborative decision-making. Based on a case study on AI-assisted breast cancer screening, two important findings can be highlighted. First, decomposition and decoupling through temporal division of action with either humans or AI dominating enable an advanced human-AI decision process to be decoupled while enabled by a foundation of shared situatedness. Second, decision-making emerges as a dynamic sensemaking process with each additional human-AI interaction evolving the decision-making process until a final decision outcome is reached.