AI Literacy, Trust and Learning: What African Students Think About AI Teaching Assistants

AI teaching assistants are entering African universities less as futuristic replacements for lecturers and more as practical tools for a familiar problem: students need timely academic support, but human support is often stretched.

A qualitative study from a South African clinical education setting shows that students welcomed AI teaching assistants for quick explanations, after-hours help and revision support, yet they remained cautious about accuracy, over-reliance and the limits of AI in clinical training. Their response was conditional acceptance: AI can support learning, but it cannot replace the human judgment, mentorship and supervised practice that clinical education requires.

Conducted by Zijing Hu at the University of Johannesburg and published in Trends in Higher Education, the study shows that responsible adoption will depend on more than simply giving students access to new tools.

Students Value AI for What It Does, Not How It Works

Students understand AI teaching assistants mainly through practical use. They know that AI can explain concepts, answer questions quickly and help them review material. But they have limited understanding of how AI systems generate responses or where those systems may fail. The study suggests that functional usefulness can drive acceptance even when conceptual understanding remains shallow. Students valued AI because it helped them solve immediate learning problems, not because they had deep knowledge of the technology behind it.

This creates both opportunity and risk. On the positive side, AI tools can lower barriers to learning by offering quick explanations and flexible support. On the risk side, students may overestimate the reliability of AI-generated answers if they are not trained to question, verify and contextualize them.

The students in the study were not uncritical users. Some said they checked AI responses against textbooks or lecturers. It points to a policy need: universities should not assume students automatically know how to use AI responsibly. AI literacy must become part of academic preparation, especially in fields where inaccurate information can have serious consequences.

Immediacy Becomes a Major Advantage

For many students, the strongest appeal of AI teaching assistants was immediacy. When lecturers were unavailable, AI offered after-hours explanations and revision support. It was particularly useful for independent learning and preparation for clinical practice.

Many institutions face large classes, limited academic staffing, uneven internet access and unequal student access to devices. In such environments, any tool that extends academic support beyond classroom hours can feel valuable. The study suggests that students' positive attitudes toward AI may be shaped by this context. AI is not necessarily attractive because the digital environment is perfect. It may be attractive because it helps compensate for gaps in support.

AI tools designed for African higher education should not assume constant connectivity, high-end devices or ideal digital infrastructure. They need to work in low-bandwidth environments, support mobile access and provide clear guidance on reliability and verification. The bigger lesson is that AI adoption should be judged by context. A teaching assistant tool that works well in a highly resourced university may not automatically suit a resource-constrained clinical programme. Successful implementation depends on alignment with local infrastructure, student needs and teaching practices.

Clinical Learning Sets a Clear Boundary for AI

Students drew a firm line between theoretical support and clinical competence. They saw AI as useful for reviewing concepts and procedures, but not as a substitute for real practice with patients or supervision by experienced professionals. Clinical education is not only about information recall. It requires judgment, ethical reasoning, communication, mentorship, embodied practice and professional accountability. AI can support knowledge acquisition, but it cannot replicate the full learning environment of clinical training.

Students repeatedly positioned AI as a supplementary tool rather than a replacement for lecturers or clinical supervisors. This guarded acceptance challenges both overenthusiastic and overly fearful views of AI. Students were open to AI, but they expected human educators to remain central.

Universities need clear rules on where AI can support learning, where human supervision is mandatory, and how students should verify AI-generated information.

The Real Challenge Is Responsible Integration

The findings point toward a balanced future for AI in African higher education. Students are willing to use AI teaching assistants when they see clear benefits. But the real challenge is responsible integration, which means universities need to build AI literacy into curricula. Students should learn not only how to use AI, but how to evaluate its outputs, recognize limitations, protect privacy and avoid overdependence. Educators also need training. AI should reduce repetitive support burdens where appropriate, but lecturers must remain active in designing learning tasks, guiding interpretation and reinforcing critical thinking.

Universities should develop guidelines for AI use in teaching, assessment, clinical preparation and academic integrity. They should also consider how AI tools handle student data, what safeguards exist against misinformation, and whether systems are accessible to students with limited connectivity or devices.

It is important to mention that the study is based on six students from one clinical module at one South African university, so the findings cannot be generalized across African higher education. It also captures student perceptions, not measured learning outcomes. The absence of educator and clinical supervisor perspectives leaves important questions unanswered. Still, the study offers a valuable early signal. It shows that students in a South African clinical learning setting see AI teaching assistants as useful, but they remain cautious about accuracy and clear about the need for human guidance.