Authors
Keywords
Abstract
Artificial intelligence-powered shopping assistants represent transformative technologies reshaping consumer behavior and retail operations through machine learning, natural language processing, and recommendation algorithms. This study empirically evaluates user perceptions across multiple dimensions of AI shopping assistant effectiveness, addressing a significant research gap in comprehensive assessment of trust, accuracy, usefulness, enjoyment, and overall experience metrics. The primary objective investigates the relationship between AI accuracy and user trust in AI-powered shopping systems to determine whether higher accuracy translates to increased trust levels. Using IBM SPSS Statistics version 27, survey data from 503 respondents were analyzed, yielding a Cronbach's alpha of 0.850, confirming high measurement reliability. The sample consisted predominantly of young to middle-aged professionals (81.1% aged 18-40), with 57.1% male respondents and high educational attainment. This python using Anderson-Darling normality tests revealed non-normal distributions across all variables, necessitating robust analytical approaches. Regression analysis produced unexpected findings: AI accuracy demonstrated a statistically significant negative relationship with user trust (β = −0.090, p = 0.008), contradicting conventional assumptions that higher accuracy increases trust. Additionally, AI trust showed no significant predictive power for recommendation intention (β = 0.034, p = 0.627, R² = 0.000). These counterintuitive results suggest complex psychological mechanisms underlying user-AI interactions that warrant further investigation. Future research should explore mediating factors such as transparency, explain ability, and user expectations that may influence the accuracy-trust relationship in AI shopping contexts.
