Artificial Intelligence Literacy, Self-Perceived Competence, and Use Among Postgraduate Medical Trainees and Faculty in Tanzania: A Multi-Institutional Cross-Sectional Study

Background Artificial intelligence (AI) is being integrated into medical education globally, yet evidence from sub-Saharan Africa remains sparse. Postgraduate medical training institutions in this region face distinctive cultural, resource and infrastructure constraints that may shape how educators and trainees engage with AI. This study examined knowledge and self-reported use of AI among postgraduate medical residents and faculty in Dar es Salaam, Tanzania. Methods A cross-sectional study was conducted across three postgraduate medical training universities in Dar es Salaam the Aga Khan University (AKU), Muhimbili University of Health and Allied Sciences (MUHAS), and Hubert Kairuki Memorial University (HKMU). A census sampling approach was used, targeting all available residents and faculty. Data was collected via an online structured questionnaire assessing AI knowledge, exposure, and self-rated competence. Knowledge scores were derived from a ten-item questionnaire and compared across groups using Welch's t-test and multivariable linear regression to determine association with knowledge scores. Results A total of 155 participants participated in the study (116 residents, 39 faculty). Mean age group was 31–40 years. Formal AI training was rare (7.1%), yet 80.0% reported using AI in teaching or learning activities. Residents used AI substantially more than faculty (65.8% vs. 14.2%). Overall objective knowledge was moderate (mean score 59.0%, SD 18.2%), with no significant difference between residents and faculty (p = 0.623). Specialty was also associated with knowledge scores, with medical and radiology specialties outperforming non-clinical specialties Self-rated knowledge at Level 5 was the only independent predictor of higher objective scores in multivariable analysis (β = 0.151; 95% CI 0.022–0.302; p = 0.049). Conclusions AI use is widespread among postgraduate trainees in Dar es Salaam despite minimal formal training. Knowledge gaps are evident, particularly around foundational AI concepts. Self-perceived competence, rather than professional role or formal training, predicted objective knowledge. These findings underscore the need for structured, contextually adapted AI literacy curricula within postgraduate medical education in Tanzania and similar settings.