The research team, led by Ayan Mohamed Jama of Somali National University, conducted a descriptive cross-sectional study during the 2025 to 2026 academic year, targeting full-time undergraduates across the university’s diverse faculties, from engineering and medicine to law and Islamic Sharia. Using Cochran’s formula with a conservative 50 percent expected proportion and a 2.5 percent margin of error, the researchers calculated a required sample of roughly 1,708 students, adjusted upward for anticipated non-response. Stratified random sampling ensured that every faculty was represented in proportion to its size, and within each stratum participants were drawn at random from official student registries. The final response rate was an impressive 98.9 percent, yielding 1,690 completed questionnaires collected over nine weeks between September and October 2025.

The instrument itself was carefully adapted to the Somali context. Knowledge was measured with ten dichotomous items covering AI concepts, definitions, and applications, while attitudes and perceptions were each assessed with ten statements on a five-point Likert scale. Three independent experts, including a higher education researcher, a public health specialist, and an ICT and AI specialist, reviewed the questionnaire for content validity, and a pretest with 5 percent of the sample refined its clarity. Because no validated AI knowledge cutoff exists for Somali students, the researchers used the mean score, approximately seven out of ten, as the empirical threshold for classifying good knowledge, a standard approach in knowledge-attitude-practice studies conducted where established benchmarks are unavailable. Reliability across subscales was confirmed with Cronbach’s alpha values at or above 0.70.

The headline results are remarkable for a country often assumed to sit at the margins of the digital economy. A full 88.5 percent of respondents had heard of artificial intelligence, and 72.7 percent demonstrated good knowledge of it, correctly identifying AI as the simulation of human intelligence by machines and recognizing everyday applications such as plagiarism detection software and personalized learning platforms. Awareness of widely used educational tools, including ChatGPT, Grammarly, and Google Translate, stood at 87.2 percent, while 84.4 percent acknowledged AI’s utility in supporting research tasks such as data analysis and literature review. The mean age of participants was 23.5 years, and digital access was surprisingly robust: 93.8 percent owned an internet-enabled smartphone and 78.4 percent had a personal computer with connectivity.

Attitudes and perceptions told a more nuanced story. Just over half of the students, 51.5 percent, expressed positive attitudes toward integrating AI into higher education, with 47.4 percent agreeing that AI can reduce academic workload and a similar share believing it can complement rather than replace teachers. Perceptions were even more favorable, with 58.1 percent holding positive views of AI’s role in enhancing accessibility and learning quality, and nearly half agreeing that AI would significantly shape the future of higher education. Yet the survey also captured a strong undercurrent of caution. More than half of respondents worried that excessive reliance on AI could erode creativity and critical thinking, 50.4 percent saw it as a potential facilitator of academic dishonesty, 52.2 percent expressed concerns about data privacy, and 53.5 percent believed Somali universities are not yet adequately prepared for AI integration.

Utilization data underscore just how embedded these tools have become. Some 82.2 percent of students reported using AI-based tools for academic purposes, and among users, 71.8 percent engaged with them daily. ChatGPT led the pack at 77.6 percent of users, followed by Grammarly at 59.5 percent and translation tools such as Google Translate and DeepL at 52.1 percent. The dominant purposes were assignment writing, cited by 94.1 percent of users, and translation or language support, cited by 77.8 percent, reflecting the practical demands of studying in a multilingual environment where English-language academic materials predominate. Among the 17.8 percent who had never used AI tools, the leading barriers were limited awareness and inadequate skills or training rather than lack of interest.

Statistical analysis revealed that adoption is not uniform. Age, faculty, and year of study were significantly associated with all four outcome variables, knowledge, attitudes, perceptions, and utilization, at p less than 0.001. Students aged 22 to 25, those in science, engineering, and health sciences faculties, and those in their second and third years showed consistently higher engagement. More unexpectedly, female students demonstrated significantly higher knowledge, more positive perceptions, and greater utilization than their male counterparts, a finding that contrasts with earlier studies reporting male predominance in AI readiness and that the authors suggest may reflect shifting gender dynamics in digital literacy among urban Somali youth.

The study’s theoretical framing helps explain these patterns. Drawing on the Technology Acceptance Model, the researchers interpret students’ positive attitudes as evidence of perceived usefulness and ease of use, the classic drivers of technology adoption. Social Cognitive Theory adds a second layer, emphasizing how knowledge, self-efficacy, and environmental factors such as institutional support and peer influence shape behavior. Viewed through this combined lens, the Somali case illustrates a distinctive adoption pathway: students are acquiring AI capabilities through informal exposure to globally accessible platforms, bypassing the structured literacy programs and governance frameworks that typically mediate adoption in high-income countries. More than 70 percent of national AI strategies worldwide belong to developed nations, and the new study shows what happens in their absence, with only 18.7 percent of students aware of any policy or guideline on AI use in education.

The international comparisons are instructive. The awareness levels observed in Mogadishu closely mirror those reported among university students in Lebanon, Saudi Arabia, and Zambia, where familiarity with AI concepts is similarly high even when formal training is scarce. The ambivalence documented in Somalia, optimism about AI’s educational potential coexisting with anxiety about integrity, privacy, and overdependence, echoes findings from Egypt, Malaysia, Sweden, and a multicenter survey across ten Arab nations. In this sense, Somali students are not outliers but participants in a global phenomenon, one in which generative AI vacancies surged ninefold between 2021 and 2024 while educational institutions everywhere have struggled to keep pace with the tools their students already carry in their pockets.

The authors argue that the policy implications are clear and urgent. Rather than attempting to restrict a technology that students have already adopted en masse, Somali universities should channel existing use through structured AI literacy programs, clear institutional guidelines on academic integrity and data protection, and the integration of AI competencies across curricula. Faculty training, investment in digital infrastructure to ensure equitable access, and national governance frameworks are equally essential. The study is not without limitations: its cross-sectional design cannot establish causation, its self-reported data are vulnerable to recall and social desirability bias, and its findings from a single Mogadishu institution may not generalize to rural areas with different levels of connectivity. Still, as the first empirical baseline of its kind for Somalia, the research delivers a vivid message: in fragile higher education systems, technological adoption is racing ahead of governance, and closing that gap thoughtfully will determine whether AI becomes a bridge to educational equity or a new source of inequality.

Subject of Research: Knowledge, attitudes, perceptions, and utilization of artificial intelligence among university students in Mogadishu, Somalia

Article Title: Knowledge, attitudes, perceptions, and utilization of artificial intelligence in higher education among university students in Mogadishu, Somalia

Article References: Jama, A. M., Tahlil, A. A., Ali, A. O., Hamud, M. O., Hassan, A. M., & Mohamed, A. S. (2026). Knowledge, attitudes, perceptions, and utilization of artificial intelligence in higher education among university students in Mogadishu, Somalia. Discover Education, 5(1), Article 1071. https://doi.org/10.1007/s44217-026-02213-2

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02213-2

Keywords: artificial intelligence, higher education, Somalia, Mogadishu, ChatGPT, AI literacy, university students, cross-sectional study, academic integrity, digital divide, Technology Acceptance Model, educational technology