AI Captain: Set Sail for Safer Shores. A Quantitative Study on AI Usage and Psychological Safety in the Top Management Team

Sammanfattning

Artificial Intelligence (AI) is increasingly integrated into strategic decision-making, yet limited research has examined how AI usage relates to psychological safety (PS) within top management teams (TMTs). This thesis investigates the relationship between AI usage and perceived PS in TMTs, and examines whether this relationship is shaped by the Unified Theory of Acceptance and Use of Technology (UTAUT) and Task-Technology Fit (TTF). Drawing on literature on TMT, UTAUT, TTF, Decision Support Systems, and PS, the study employs a quantitative cross-sectional survey design based on 198 respondents across 50 Sweden-based TMTs. The data were analyzed using Ordinary Least Squares regression, moderation analysis, and supplementary polynomial regression with Response Surface Analysis (RSA). The findings show a positive but weak direct relationship between AI usage and PS, contradicting the initial expectation that AI usage would reduce PS. This suggests that AI usage alone is not necessarily harmful, but explains limited PS variation. All UTAUT constructs positively moderated this relationship and showed highly similar empirical patterns suggesting that they functioned as related expressions of a broader AI adoption climate. Across these constructs, AI usage is more positively associated with PS when AI tools are perceived as useful, easy to use, socially supported, and institutionally enabled. TTF shows a different pattern: it is positively associated with PS, but does not significantly moderate the AI usage- PS relationship. Supplementary RSA reveals divergent alignment patterns. UTAUT alignment shows increasingly favorable PS outcomes when AI usage and adoption conditions rise together, whereas TTF alignment appears beneficial but bounded, with diminishing returns emerging when AI usage continues increasing under already high TTF conditions. The thesis contributes by showing that AI-supported strategic decision-making depends less on AI usage alone than on the surrounding adoption and fit conditions.