Distributed Cognition and Human-AI Delegation in Knowledge Work

This ongoing study explores how human-AI delegation manifests in knowledge-intensive work through ethnographic observation of radiologists using AI-diagnostic tools. We ask: How does task delegation between humans and AI take place in this dynamic knowledge workplace? Using the framework of Distributed Cognition, we analyze how diagnostic process of radiology is collaboratively performed across human-machine configurations. Unlike static and stiff systems of the past, AI introduces indeterminacy and doubts, challenging established assumptions about the process of knowing and trusting. We find that epistemic trust – the confidence in the reliability, truthfulness, and validity of knowledge claims, is neither a given nor a prerequisite, but a dynamically negotiated practice that enables delegation. Epistemic trust helps calibrate human judgment, mitigate automation bias and aversion, and in process sustain agency in high-stakes environments. Current findings point to advancement of distributed knowledge in sociotechnical systems and reveal how trust shapes cognitive delegation in AI systems.