Preliminary Validation of the Italian Version of the Artificially Intelligent Device Use Acceptance (AIDUA-IT) Scale: Cross-Cultural Adaptation and Psychometric Evaluation

TL;DR

This study validates an Italian version of a scale to measure acceptance of AI devices, showing it is reliable and valid for use in Italy.

Contribution

The study provides the first validated Italian version of the AIDUA scale for assessing AI device acceptance.

Findings

The AIDUA-IT scale showed excellent fit and strong psychometric properties in structural validity and internal consistency.

Convergent and discriminant validity, as well as test–retest reliability, were supported across subscales.

The AIDUA-IT is a reliable instrument for assessing AI acceptance in Italian-speaking populations.

Abstract

Background: Artificial intelligence (AI) is increasingly integrated into healthcare and public services, making user acceptance a key prerequisite for safe and effective implementation. The Artificially Intelligent Device Use Acceptance (AIDUA) model provides a multidimensional framework for evaluating acceptance of intelligent systems, yet no validated Italian instrument is currently available. Objectives: This study aimed to translate, culturally adapt, and preliminarily validate the Italian version of the AIDUA scale (AIDUA-IT) following COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) and Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) recommendations. Methods: A two-phase cross-sectional design was used. Phase one included forward–backward translation, expert review (n = 7), and cognitive debriefing (n = 8). Phase two assessed structural validity, internal consistency, convergent and discriminant validity, and short-term test–retest reliability in a convenience sample of Italian-speaking adults (N = 140), with a subsample completing the test–retest assessment (n = 32). Results: The hypothesized eight-factor measurement model demonstrated excellent fit (Comparative Fit Index [CFI] = 0.984; Tucker–Lewis Index [TLI] = 0.981; Root Mean Square Error of Approximation [RMSEA] = 0.041; Standardized Root Mean Square Residual [SRMR] = 0.056), with strong standardized loadings (β range: 0.64–0.96) and good internal consistency (Cronbach’s α and McDonald’s ω range: 0.82–0.90). Convergent and discriminant validity were supported, and test–retest reliability was good to excellent across subscales (Intraclass Correlation Coefficient [ICC] range: 0.81–0.90). Conclusions: These findings provide initial evidence that the AIDUA-IT is a reliable and valid instrument for assessing acceptance of AI-enabled services in Italy. Further validation in larger and more diverse samples is recommended.

Figures