Adaptation of Artificial Intelligence Attitude Scale (AIAS-4) into Turkish: a validity and reliability study

Abstract

The integration of Artificial Intelligence (AI) into all areas of our lives has been creating substantial changes. There certainly is an ever-growing need to assess the way public attitudes towards AI have taken shape. This study aims to adapt the Artificial Intelligence Attitude Scale (AIAS-4) (Grassini, 2023) into Turkish, where there are no measures that can be used in different contexts and by different organizations to assess the overall attitude towards AI and test its validity and reliability. The scale adaptation was conducted in two phases. The study was carried out with the participation of 422 undergraduate students during the 2023-2024 academic year. To verify the construct validity of the scale, exploratory factor analysis was conducted with 221 undergraduate students. With another group of 201 undergraduate students, confirmatory factor analysis, test-retest reliability and measurement invariance tests were computed. As in the original scale, results from exploratory and confirmatory factor analyses justified the single-factor structure of the scale with four items explaining 63.03% of the total variance, and item loadings were in the range of 0.70-0.89. The model fit values were in a good fit range (CFI: 0.998, TLI: 0.993, RMSEA: 0.048). Also, the model was strictly invariant between genders. The scale demonstrated good internal consistency in two studies, with Cronbach's alpha of 0.86 and 0.86; and McDonald's omega of 0.87 and 0.86. The results indicated that all items had discriminative power and therefore, the Turkish version of the scale proves to be valid and reliable and thus could be used by researchers in the Turkish context.