BMC Medical Education Education

Self-reported artificial intelligence familiarity, use, and attitudes among students, faculty, and staff at a medical college in the United Arab Emirates: a cross-sectional survey

Ghenimi, Nadirah, Govender, Romona D., Almarzooqi, Falah Mohammed, Adam, Hiba, Shaban, Sami, Al-Shamsi, Saif
Springer Nature · 2026-11-20 · DOI 10.1186/s12909-026-10921-4

Abstract

Background Artificial intelligence (AI) is increasingly used in healthcare and medical education. We described self-reported AI familiarity, use, and attitudes among students, faculty, and staff at a medical college in the United Arab Emirates to inform local training. Methods We conducted a cross-sectional online survey from August to October 2025. Study-specific composite summary scores ranged from 0 to 100. Linear models adjusted for respondent group, sex, and age, with HC3 standard errors. Sensitivity analyses examined AI use scoring, bootstrap confidence intervals, and restricted-age group comparisons. Results After four blank records were excluded, 375 respondents remained; the estimated participation proportion was 34.7%. At least one use of a generative AI tool during the previous three months was reported by 96.8% of respondents, whereas 78.1% reported AI use specifically in academic studies or professional practice. Formal AI training was reported by 10.8%. Mean AI familiarity, AI use, and AI attitude scores were 61.6, 62.9, and 72.1. Adjusted student-versus-faculty/staff differences were 1.91 points for AI familiarity (95% confidence interval [CI] -6.99 to 10.81), -2.87 for AI use (95% CI -23.17 to 17.43), and 2.32 for AI attitude (95% CI -5.43 to 10.07). Because age overlap between students and faculty/staff was limited, these adjusted group estimates may partly rely on extrapolation and should not be interpreted as age-matched comparisons. Average linear-model age associations indicated lower AI familiarity and higher AI attitude scores. The adjusted male-versus-female AI use estimate was uncertain (-7.61, 95% CI -16.14 to 0.92). The higher adjusted AI attitude score among male respondents was exploratory and requires caution because no multiplicity adjustment was applied. Training-stage comparisons were exploratory and showed no consistent pattern. The restricted-age analysis included 59 respondents and had wide confidence intervals. Conclusions Any recent generative AI use was common, while formal training was uncommon. These findings may guide local teaching on responsible AI use, checking AI-generated information, and maintaining human oversight. Self-report, one institution, and study-specific scores that were not formally validated limit interpretation. Future studies should validate the measures and include objective assessments of AI competency.

Read at publisher →