Dental students’ perceptions and use of artificial intelligence in dental education in Bulgaria: a comparative cross-sectional survey of two independent cohorts in 2025 and 2026

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Abstract Background Artificial intelligence (AI) is increasingly influencing dental education, clinical decision-making, diagnostics, and treatment planning. However, cohort-level differences in AI-related perceptions and use among dental students from consecutive academic years remain insufficiently explored, particularly among international students enrolled in English-language dental programs. Objective This exploratory study aimed to describe and compare AI-related perceptions, usage patterns, perceived benefits, concerns, and educational expectations between two independent cohorts of third-year undergraduate dental students surveyed in 2025 and 2026 using an identical questionnaire. Methods Two anonymous cross-sectional questionnaire surveys were conducted among independent cohorts of third-year undergraduate dental students enrolled in an English-language dental medicine program. The same questionnaire was administered in both years. The 2025 cohort included 109 respondents, and the 2026 cohort included 92 respondents. Collected demographic variables included gender, age, and country of origin. Categorical variables were compared using chi-square tests, and ordinal responses were analyzed using Mann–Whitney U tests. Because no primary outcome was prespecified, all inferential analyses were considered exploratory. The Holm step-down procedure was applied across the reported inferential tests to address multiple testing, and statistical interpretation was based on Holm-adjusted P values. Results A total of 201 students participated. A higher proportion of the 2026 cohort reported frequent self-reported AI use from 22 of 109 students (20.2%) in 2025 to 27 of 92 students (29.3%) in 2026, while “rarely or never” responses decreased from 35 of 109 students (32.1%) to 19 of 92 students (20.7%). In unadjusted analyses, AI-use frequency differed between the cohorts (Mann–Whitney U = 4243.5; unadjusted P =.042; r =.14), and the distribution of perceived benefits also differed (χ²₃=8.27; unadjusted P =.041; Cramér’s V=0.20). However, both effect sizes were small, and neither comparison remained statistically significant after Holm correction (adjusted P ≈.82 for both). Familiarity with regulations and legal guidelines remained limited: 68 of 109 students (62.4%) in 2025 and 46 of 92 students (50%) in 2026 reported being “not familiar” with such guidelines. Perceived institutional preparedness remained low, with 69 of 109 students (63.3%) in 2025 and 62 of 92 students (67%) in 2026 reporting that their institution did not provide sufficient AI-related training. Conclusions Descriptively, the 2026 cohort reported more frequent AI use and more often identified treatment planning as the principal perceived benefit of AI. These small cohort-level differences did not remain statistically significant after adjustment for multiple testing and should be interpreted as exploratory rather than confirmatory evidence of temporal change. Ethical, legal, and institutional training gaps were evident in both cohorts, supporting the need for structured AI literacy, regulatory awareness, and clinically oriented AI training within dental curricula in Bulgaria.