General Artificial Intelligence Attitude Scale (GAIA-15) – How to measure attitude toward an ever-changing object

The influence of Artificial Intelligence (AI) continues to grow and attitude toward AI is becoming an increasingly important scientific construct. However, to study AI attitude systematically, we must first operationalize it. We introduce the General Artificial Intelligence Attitude Scale (GAIA-15), a 15-item measure that assesses affective, behavioral, and cognitive aspects of AI attitude. For scale development, we conducted a review of existing scales and created new items. Then we used item response theory to select and integrate the best items into our GAIA-15 scale. Moreover, we provide the General Artificial Intelligence Attitude Item Pool (GAIA-I) from which researchers can draw customized scales. Providing the GAIA-I addresses a challenge linked to the assessment of AI attitude: unlike most attitude objects, AI is rapidly evolving and a measure of attitude toward AI must be adaptable to future developments of AI. The approach of an IRT-based item pool grants this flexibility. Using an online questionnaire, we validated both GAIA tools with a sample of 302 participants. We found evidence for convergent validity (alignment with established AI attitude scales), discriminant validity (distinction from the constructs “AI anxiety” and “innovativeness”), as well as predictive validity (relation to AI use in a follow-up survey). • Development of the assessment tools is based on Item Response Theory (IRT). • The instruments assess affective, behavioral and cognitive aspects of AI attitudes. • GAIA-15 and GAIA-I show good reliability, strong validity and predict future AI use. • The GAIA-15 scale outperforms existing AI attitude scales in IRT model fit. • The GAIA-I item pool allows future-proof customization of AI attitude scales.