Purpose:
This study investigated the relationship between pre-service science teachers’ epistemic beliefs and their attitudes towards generative artificial intelligence (AI) tools. Additionally, it examined the moderating effects of demographic variables on this relationship.

Method:
A correlational research design was employed. The participants were 299 pre-service science teachers enrolled at a uniersity’s physics (n=75), chemistry (n=74), biology (n=75), and science education (n=75) programs. Data were collected using Schommer's Epistemological Beliefs Questionnaire and the Generative Artificial Intelligence Acceptance Scale. Reliability analysis revealed that the Scientific Epistemological Beliefs Scale demonstrated good reliability (Cronbach's α = .813, standardized α = .834), while the Generative Artificial Intelligence Acceptance Scale showed acceptable reliability (Cronbach's α = .767, standardized α = .891) according to George and Mallery's (2003) criteria. Spearman correlation, multiple regression, and moderation analyses were employed to analyze data.

Findings:
The results revealed significant positive correlations between dimensions of epistemic beliefs and AI acceptance. The strongest correlation was observed between beliefs about the structure of knowledge and AI acceptance (r=0.492, p<0.001). Multiple regression analysis showed that beliefs about the source of knowledge (β=.254), the nature of knowledge (β=.204), and the learning process (β=.158) accounted for 15.1% of the variance in AI acceptance. Demographic variables demonstrated significant moderating effects: gender had a particularly strong moderating role in the dimension of certainty of knowledge. A strong positive relationship was found among female participants while no significant relationship was observed among males. The analysis on academic level indicated that the impact of epistemic beliefs on AI acceptance progressively increased over the academic years.

Conclusion:
The findings suggest that epistemic beliefs provide a critical cognitive foundation for the acceptance of AI technologies among pre-service science teachers. However, these relationships are significantly moderated by demographic characteristics. Gender-specific epistemic processing patterns highlight the need for differentiated approaches in technology integration training. The results emphasize the importance of integrating epistemic belief development with technology integration training in teacher education programs and highlight the need for theoretically grounded and demographically responsive approaches to effectively integrate AI technologies in science education.

Practical Implications:
The findings indicated that teacher education programs should adopt approaches that recognize epistemic diversity and consider gender, academic level, and disciplinary differences. Developing epistemic awareness among teacher candidates is essential for the responsible and effective use of AI technologies in education.

Keywords: Epistemic beliefs, artificial intelligence acceptance, pre-service science teachers, technology integration, demographic moderation effects.