Přehled o publikaci

This study examines how perceived AI system reliability shapes knowledge identity (KI) processes in organisations by influencing how knowledge is presented, transferred, and reproduced. Prior work on AI trustworthiness has largely emphasised technical performance, while underspecifying reliability as a socio-cognitive judgement enacted through human–AI interaction. Drawing on socio-technical systems theory and identity-based knowledge management, a mixed-method design was employed in an aviation maintenance context. First, grounded interviews with domain experts identified five human-centred reliability dimensions (accuracy, user trust, explainability, consistency, and responsiveness). Second, the Analytic Hierarchy Process was used to prioritise these dimensions and inform a composite system reliability score (SRS). Third, hypotheses were tested using PLS-SEM on survey data from 116 knowledge workers. Results show that perceived AI reliability positively affects all three KI processes, with accuracy and user trust exerting the strongest influence. The study contributes by (i) conceptualising AI reliability as a socio-cognitive construct central to KI formation, (ii) operationalising a weighted SRS to support evaluation and improvement of AIenabled knowledge management (KM) systems, and (iii) providing actionable design implications for aligning AI tools with organisational sensemaking and knowledge continuity. Future research should validate the model across industries and integrate objective system metrics with perception-based reliability measures.