Beyond Positive Response Rates: Capturing Information Richness in Workplace AI Acceptance Using Belief Structure TOPSIS

This study applies the Belief Structure TOPSIS (B-TOPSIS) method to analyse cross-country attitudes toward AI-driven workplace practices across the EU27. The proposed approach preserves the full distribution of survey responses, explicitly incorporates uncertainty, and evaluates alternatives based on their distance from ideal and anti-ideal belief structures. Using data from Special Eurobarometer 554, we construct individual B-TOPSIS indexes for eight AI-related workplace applications and an aggregated B-TOPSIS index capturing overall acceptance. The results reveal systematic cross-country differentiation. Activities such as gathering applicant information, allocating work, and processing employee data generally receive moderate acceptance. Safety-focused applications are widely supported, whereas ethically sensitive practices, such as employee monitoring and automated dismissal, face low acceptance. Additionally, sensitivity analysis based on Monte Carlo simulation and stochastic dominance demonstrates that the obtained rankings remain highly stable under alternative assumptions regarding utility functions, confirming the robustness of the proposed framework. A comparison with rankings derived from total positive responses, commonly used in EU reports, shows that although the two approaches are strongly correlated, they are not interchangeable. By retaining the complete response structure, the proposed method captures differences in response intensity that are obscured by conventional summary measures. The findings highlight the multidimensional and conditional nature of workplace AI acceptance in the EU and demonstrate the value of belief-structure-based approach for analysing survey data.