The rapid advancement and widespread adoption of artificial intelligence (AI) technologies have profoundly influenced social production and daily life. Public attitudes toward AI directly shape individuals' willingness to use and adopt the technology, which in turn affects whether AI can develop in a responsible and sustainable manner. Currently, the most representative and widely used instrument for measuring attitudes toward AI is the General Attitudes towards Artificial Intelligence Scale (GAAIS). However, its psychometric properties have not yet been examined among Chinese university students. As a pioneer group deeply engaged with AI technology, university students' attitudes toward AI are crucial to their technological adaptation and literacy development. Therefore, the present study aimed to revise the Chinese version of the GAAIS and examine its reliability, validity, and measurement invariance across gender, independent samples, and time among Chinese university students, thereby providing a reliable measurement tool for domestic research.

Using convenience sampling, university students were recruited from Sichuan, Shandong, Anhui, Guizhou and other provinces, yielding 1970 valid participants. Four independent samples were collected: Sample 1 (n=631) was used for item analysis and exploratory factor analysis; Sample 2 (n=714) was used for confirmatory factor analysis and reliability analysis; Sample 3 (n=625) was used for criterion-related validity tests and cross-validation of the factor structure via confirmatory factor analysis; the combination of Sample 1 and Sample 2 (n=1345) was used for measurement invariance tests across gender and for gender differences; the combination of Sample 2 and Sample 3 (n=1339) was used for cross-sample measurement invariance tests; Sample 4 (n=185) was randomly selected from Sample 2 after a four-week interval for test-retest reliability analysis and longitudinal measurement invariance tests. The Brief Artificial Intelligence Attitude Scale, Daily Artificial Intelligence Usage Scale, Consideration of Future Consequences Scale, the Openness to Experience Subscale, and the Life Orientation Test were employed as criterion measures. Data were analyzed using SPSS 25.0 and Mplus 8.3.

Item analysis indicated good discrimination for all items. Exploratory factor analysis revealed a 16-item Chinese version of the GAAIS with two dimensions: positive attitudes (9 items) and negative attitudes (7 items), accounting for 54.26% of the total variance, with factor loadings ranging from 0.65 to 0.80. Confirmatory factor analyses using both Sample 2 and Sample 3 supported a good fit for the two-factor model (Sample 2: χ²/df=1.91, RMSEA=0.04, CFI=0.97, TLI=0.97, SRMR=0.04; Sample 3: χ²/df=1.85, RMSEA=0.04, CFI=0.97, TLI=0.97, SRMR=0.04). Criterion-related validity analysis showed that the GAAIS total scores and positive attitudes were significantly positively correlated with brief AI attitude, AI usage frequency, future orientation, openness to experience, and life orientation, and significantly negatively correlated with current orientation (all ps<0.001). In contrast, negative attitudes showed the opposite pattern. Reliability analysis indicated that Cronbach's α coefficients for the total scale and its dimensions ranged from 0.86 to 0.89, split-half reliability coefficients ranged from 0.85 to 0.88, and test-retest reliability coefficients ranged from 0.67 to 0.76. Multi-group confirmatory factor analysis demonstrated strict measurement invariance across both gender and independent samples (ΔCFI≤0.01, ΔTLI≤0.01, ΔRMSEA≤0.015). Longitudinal measurement invariance tests showed partial strong invariance across the 4-week interval (ΔCFI=-0.008, ΔTLI=-0.006, ΔRMSEA=0.005). Gender difference tests revealed that males scored significantly higher than females on the positive attitudes dimension (t=2.48, p=0.013, Cohen's d=0.14).

The revised Chinese version of the GAAIS consists of 16 items across two dimensions (positive and negative attitudes) and demonstrates good reliability and validity among Chinese university students. Measurement invariance analyses showed strict invariance across gender and independent samples, and partial strong invariance across time. This two-factor structure is clear and stable, with all psychometric indicators meeting acceptable standards. Therefore, the scale can serve as a reliable instrument for assessing Chinese university students' general attitudes toward AI.

This revision of the GAAIS provides a standardized measurement foundation for domestic research on AI attitudes. Future studies can utilize this instrument to investigate the characteristics and influencing factors of university students' AI attitudes, explore their relationships with variables such as academic performance, career planning, and technology usage behaviors, and evaluate the effectiveness of AI-related courses or interventions. This scale also provides empirical evidence for AI literacy education and the promotion of technology acceptance in higher education institutions.