Artificial Intelligence, Large Language Models, and Machine Learning in African Health Professions Education: A Systematic Review of Adoption, Literacy, Institutional Feasibility, and Ethical Perceptions

In the classrooms and hospitals of Africa, a quiet transformation is underway. It is not driven by new textbooks or updated syllabi, but by a technology that has swept across the globe in recent years: artificial intelligence. This is not the distant, robotic future of science fiction, but a set of tools already in the hands of students and teachers. Among these tools are large language models, which are computer programs capable of understanding and generating human language, and machine learning, a method where computers learn from data to make predictions or decisions. For medical, nursing, and pharmacy students, these tools promise to be as essential as stethoscopes or microscopes in the coming decades. They can help diagnose diseases, organize research, and even draft patient notes. Yet, while high-income nations have begun to build formal structures to teach how to use these tools safely and effectively, the situation across the African continent remains a complex mix of eager adoption and significant uncertainty. The question facing educators and policymakers is no longer whether these students will use artificial intelligence, but how they are using it, what they truly understand about it, and whether their institutions are ready to guide them.

A new systematic review, a type of research that gathers and analyzes many different studies to find the big picture, has set out to answer these questions specifically for Africa. The researchers, a team of scholars from universities in Nigeria, the United Kingdom, and the United States, looked at nineteen separate studies involving thousands of health professions students and educators across the continent. They searched for evidence on how familiar these students were with artificial intelligence, whether they had received any formal training, what tools they preferred, and what ethical worries kept them up at night. The goal was to move beyond scattered reports and build a clear understanding of the reality on the ground, from the bustling medical schools of Nigeria and Egypt to the universities of Ghana and Sudan.

What the researchers found was a striking contradiction, a gap between what students think they know and what they actually know. Across the studies, awareness of artificial intelligence was high. In some groups, nearly all students knew what these tools were and had heard of them. However, this familiarity did not translate into competence. When students were asked to rate their own skills, many felt confident, but when tested with objective questions, their knowledge was often quite low. It is as if a student could recognize the name of a famous instrument but could not play a single note. This "knowledge-perception paradox" suggests that students are learning about these tools on their own, through curiosity and peer conversation, rather than through structured lessons. In one large survey of medical students from forty-eight countries, fewer than one percent reported receiving substantial formal education on artificial intelligence, even though most claimed to have at least a moderate understanding of the subject.

The tool driving this informal revolution is overwhelmingly ChatGPT. When the researchers looked at which specific programs students were using, this single application dominated the landscape, far outpacing other competitors. Students turned to it for practical reasons: it was easy to use, it saved time, and their friends recommended it. They used it to write essays, prepare for exams, search for medical literature, and even to draft research papers. This adoption happened largely outside the walls of the classroom. There was little evidence of universities integrating these tools into their official curricula or providing guidance on how to use them responsibly. Instead, the students were navigating this new digital terrain alone, driven by individual initiative rather than institutional direction.

This lack of guidance is compounded by a harsh reality of infrastructure. The review highlighted that for many students, the basic conditions needed to use these tools are not guaranteed. Unreliable electricity, expensive data plans, and spotty internet connections were frequently cited as major barriers. In some places, the lack of a stable power supply or the high cost of internet access meant that even when students wanted to use these advanced tools, they simply could not do so consistently. Yet, the story is not entirely one of limitation. In a few instances, the researchers found that artificial intelligence was being used to solve these very infrastructure problems. For example, one study showed how AI could generate backgrounds for educational videos, removing the need for expensive physical studios and green screens. In this way, the technology sometimes acted as a workaround for the very resource constraints that usually hinder its use.

Perhaps the most consistent finding, however, was the deep concern students felt about the ethical implications of these tools. Despite their willingness to use artificial intelligence, students were acutely aware of the risks. They worried about academic dishonesty, the possibility of relying too much on the computer and losing their own critical thinking skills, and the danger of the tools providing incorrect or made-up medical information. These fears were not just abstract; students admitted to using AI to complete exams and to check their answers before submitting them. They also expressed anxiety about the future of their professions, wondering if these tools would eventually replace human doctors or dehumanize patient care. Interestingly, the more students knew about artificial intelligence, the more worried they often became about these risks, suggesting that familiarity brings a sharper awareness of the potential dangers.

The review concludes that while African health professions students are rapidly embracing artificial intelligence, they are doing so in a vacuum of formal support. The technology is being adopted informally, driven by student curiosity and the sheer accessibility of tools like ChatGPT, but it is moving faster than the institutions can adapt. There is a clear need for structured education that goes beyond just teaching students how to click a button. The researchers argue that universities must step in to provide formal training that covers not only technical skills but also the ethical dimensions of using these tools. This includes teaching students how to spot errors, how to maintain academic integrity, and how to use artificial intelligence as a partner in learning rather than a replacement for their own judgment. Without this guidance, the gap between the students' growing reliance on technology and their institutions' ability to prepare them will only widen, leaving a generation of future healthcare workers navigating a complex digital landscape without a map.