Logo: to the web site of the Swedish Defence University

This pilot study examines whether trust in AI-supported intelligence analysis can be operationalised and analysed through experimental vignette design in a Swedish military context. Twenty-five officers with military intelligence analysis experience were randomly assigned to one of four vignettes manipulating analytic tradecraft dimensions (transparency and alternative hypotheses) in AI-generated assessments. Situational trust (S-TIAS), behavioural intention (BI), and dispositional trust (HTP and MTP) were measured. The instruments demonstrated high internal consistency, and situational trust correlated strongly with behavioural intention, indicating that trust functions as a context-specific mechanism in this setting rather than a general attitude toward technology. The tradecraft manipulations produced small and non-significant effects, suggesting limited detectability in a small pilot sample. Qualitative responses nevertheless highlighted transparency and structured reasoning as central to professional trust assessments. The study demonstrates methodological feasibility in a security-sensitive setting while identifying design constraints for future, larger-scale investigations of AI integration in military decision-making under uncertainty.