Work

Align, then assign: Comparing Active vs. Passive Conversational Assistants on Clinician Workload and Agency During Care Quality Analysis

Year: 2026

Type: conference-paper

Abstract: Conversational AIs (CAIs) can assist clinicians when analysing their hospitals' care quality by automating the execution of analysis based on users' commands. However, CAIs have different levels of participation such as passive; only carrying commands, and active; also offering suggestions. While active CAIs can reduce workload, they also risk automation bias and loss of agency. The exact needs and balancing of this trade-off remains unclear in current research on CAI design for clinical use. Therefore, we carried out four studies consisting of two interview rounds on the design of CAI behaviour and two studies evaluating CAIs including a comparison between active and passive participation in a care quality task. The active CAI reduced workload and led to higher success rates in the task while increasing agency. Furthermore, we explore concrete needs and future directions for CAI design to better incorporate them into care quality improvement workflows.

Source: VBN Forskningsportal (Aalborg Universitet)

DOI: 10.1145/3843148.3843160

Authors: Ziadeh, Hamzah; id_orcid 0000-0001-9288-6520, Palaiodimou, Lina, Tsivgoulis, Georgios, Antova, Izabel, Taskov, Dimitar, Quinn, Terry, Mikulik, Robert, Knoche, Hendrik; id_orcid 0000-0003-3950-8453

Institutions: Aalborg University, National and Kapodistrian University of Athens, University General Hospital Attikon, Multiprofile Hospital for Active Treatment in Neurology and Psychiatry St. Naum, University of Glasgow, Masaryk University

Language: English

Topic: Artificial Intelligence in Healthcare and Education

Subfield: Health Informatics

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: green

OpenAlex ID: W7204257501

FWCI:

Cites:

Cited by:

Related to: 10