Evaluating Egyptian citizens’ perception toward introducing voice assistant technology as a means of improving public service delivery: utilizing machine learning as an additional perception’s predictor

Abstract

Purpose: This research assesses the inclination of Egyptian citizens toward embracing Voice Assistant Technology (VAT) to deliver public services based on the functioning of perceived usefulness, ease of use, trust, and perceived risk. This also examines the possibility of using machine learning (ML) models to forecast adoption behavior.

Design/methodology/approach: A mixed-method design was applied, supplementing survey data from 398 participants with qualitative analyses of expert interviews. An extended Technology Acceptance Model (TAM) incorporating trustworthiness and perceived risk was employed. Additionally, ten (ML) algorithms were applied to predict acceptance by citizens.

Findings: Helpful conclusions were reached, the most helpful being that usefulness, ease of use, and trust highly and positively affect (VAT) acceptance while perceived risk highly and negatively affects VAT acceptance. (ML) analysis validated these findings with Stochastic Gradient Descent (71.9% accuracy) and Ridge Regression (70.9%) as the best predictors, yet Decision Tree was poor (49.3%). These conclusions indicate that risk perceptions need to be addressed and trust enhanced to facilitate VAT adoption in developing-country contexts.

Originality/value: This paper contributes to the field by extending TAM with trust and risk factors and adding ML predictive modeling to public administration studies. The results provide policy practitioners and technologists with actionable advice on how to incentivize AI-enabled public service delivery via citizen-focused, trust-building approaches.