Nigerian Undergraduates Turn to AI, Reshaping How They Learn, Study Warns

Artificial intelligence has swept through university campuses worldwide since 2023, but a new qualitative study from Nigeria offers one of the most detailed windows yet into how this transformation actually feels from the student’s side of the lecture hall. Published in Discover Education, the research by Nweke Gerald Chigozie of the University of Nigeria, Nsukka, interviewed twelve undergraduates drawn from one federal university in each of Nigeria’s six geopolitical zones, probing how they first encountered AI, how the technology has rewired their academic routines, and where it has let them down. The findings paint a picture of a generation learning faster, working more independently, and, in some cases, quietly outsourcing the very thinking their degrees are meant to train.

The study rests on two theoretical pillars. The first is the Technology Acceptance Model, developed by Fred Davis in 1986, which holds that people adopt a technology when they judge it useful and easy to use. In this framing, Nigerian students embrace AI because they perceive it as an instrument that improves access to information, simplifies difficult concepts, and supports learning activities, and that acceptance then shapes how they interact with resources, seek information, and complete tasks. The second lens, Barry Zimmerman’s Self-Regulated Learning theory from 1989, treats students as active agents who plan, monitor, and evaluate their own learning. Crucially, the theory predicts that AI’s impact depends on regulation: used deliberately to clarify concepts and gather tailored resources, it can support autonomy; used as a substitute for cognitive effort, it can quietly hollow out the learning it appears to accelerate.

Methodologically, the research is exploratory and deliberately small. One federal university was purposively chosen from each geopolitical zone: the University of Ibadan in the South-West, the University of Nigeria, Nsukka in the South-East, the University of Port Harcourt in the South-South, the University of Maiduguri in the North-East, the University of Abuja in the North-Central, and Ahmadu Bello University in the North-West. From each institution, one male and one female undergraduate were selected, yielding twelve participants aged 20 to 29 who already used AI tools for academic work. Semi-structured in-depth interviews lasting roughly 40 to 45 minutes were conducted by telephone or Zoom, audio-recorded with consent, transcribed verbatim, and coded inductively with NVivo 12 software. The interview guide was vetted by experts in qualitative research and educational technology and pilot-tested before the main data collection, with two professors independently reviewing the coding to strengthen analytical rigour.

One of the study’s most striking findings concerns where awareness begins. Not one participant credited formal university instruction as their gateway to AI. Instead, they described encountering the technology through social media platforms such as YouTube and Twitter, through classmates who demonstrated tools in action, and, less often, through lecturers who recommended AI when students struggled with course content. One male student recalled stumbling onto a YouTube platform two years earlier that framed AI as a way of easing academic stress; a female student described a Twitter conversation that transformed her view of AI from a threat into a learning aid. This informal diffusion matters because awareness acquired through social media and peer networks shapes expectations before any classroom policy does. The finding echoes earlier Nigerian surveys in which peers, friends, and social media far outstripped official channels as sources of first exposure to tools like ChatGPT, suggesting universities arrive late to a conversation already well underway among students.

Once adopted, AI visibly restructured four dimensions of academic behaviour. Students reported dramatically faster access to materials, with some describing the library as nearly obsolete for routine information needs. One participant explained that AI lets him skip less important details and extract the main points of a course in seconds, replacing weeks of textbook reading. Second, assignment completion became markedly quicker and less effortful, easing deadline pressure across disciplines. Third, and more positively, the technology facilitated genuinely independent study and research, with students exploring topics beyond lecture content and at their own pace, free from the scheduling constraints of lecturer availability. Fourth, participants prized AI’s ability to tailor explanations, prompting chatbots to rephrase complex concepts at their preferred level of comprehension and to answer follow-up questions in ways standard textbooks cannot. Taken together, these accounts describe a shift from hunting for information toward evaluating and interrogating material that arrives pre-synthesised.

Yet the same interviews surfaced a darker ledger. Students described AI-generated information that was inaccurate, outdated, or misleadingly confident, sometimes distorting concepts they did not yet understand well enough to catch the errors. One participant reported receiving a misinterpretation of an idea she had queried; another described outdated facts and content errors in an AI-assisted assignment, concluding that students should never take AI answers at face value. A second cluster of problems involved AI literacy itself: several students admitted that their early prompts produced poor results simply because they lacked the skills to formulate effective queries or assess outputs. A third was infrastructural. Unreliable campus Wi-Fi and unstable connectivity repeatedly disrupted access to AI platforms, with one student noting the network was usable mainly at night, raising equity concerns for students who cannot afford private data plans.

The most sobering theme, however, was overdependence. Participants volunteered candid admissions that habitual AI use had dulled their willingness to think independently. One female student said she now felt lazy thinking on her own, having used AI for assignments, term papers, and exam preparation. A male participant went further, describing AI as a substitute for his thinking mind that had slowed his cognitive strength and prompting him to declare it time to regulate his own use. These confessions align with broader experimental evidence that heavy reliance on tools like ChatGPT can reduce students’ mental effort, and with the self-regulated learning framework’s warning that AI becomes harmful precisely when it replaces, rather than scaffolds, the cognitive work of learning. The study frames this not as an inherent property of the technology but as a failure of regulation, one that universities can address.

Participants themselves proposed remedies, and the study organises them into four strategies. First, students should be trained to critically verify AI outputs against credible academic sources before citing or using them, sharpening analytical habits rather than eroding them. Second, institutions should provide formal training on ethical AI use, covering plagiarism, bias, attribution, and academic integrity, so students stop perceiving AI as an ultimate source of learning that shortchanges individual effort. Third, AI literacy should be embedded directly into the university curriculum, with several participants arguing it should be compulsory, so that every graduate learns prompt formulation, output evaluation, and the technology’s limits. Fourth, governments and universities must invest in reliable digital infrastructure, including stable electricity and internet connectivity, because AI-supported learning remains inequitable without it.

The practical implications reach into assessment design as well. The finding that AI lets students finish assignments with little effort suggests that traditional take-home essays are increasingly poor measures of learning. The study points toward process-based assignments, in-class activities, oral defences, and reflective components that require students to critique and justify AI-assisted responses rather than submit them unexamined. Lecturers, identified as gatekeepers whose recommendations confer academic legitimacy on AI tools, are urged to move beyond banning or blessing the technology and instead model legitimate uses, such as brainstorming, clarification, and feedback, while holding students accountable for argumentation and final decisions. Peer networks, already the dominant diffusion channel, could be harnessed through student-led workshops and AI learning communities to spread responsible practice rather than shortcuts.

The author is careful to note the limits of what twelve interviews can establish: the findings are exploratory insights rather than generalisable conclusions, and future work should include larger samples, educators’ perspectives, and longitudinal or mixed-methods designs tracking how AI-mediated learning evolves. Still, the study’s significance lies in its refusal to reduce the AI question to adoption statistics. Across Germany, the Nordic countries, the United States, Ghana, South Africa, and Kenya, surveys show students flooding into AI use, but this Nigerian investigation reveals the behavioural mechanics underneath: informal awareness channels precede formal policy, efficiency gains coexist with cognitive offloading, and the educational value of AI ultimately hinges on each student’s capacity to regulate their own engagement. In that sense, the study’s central message travels well beyond Nigeria’s borders. The technology is not inherently a tutor or a crutch; it becomes whichever one the user’s self-discipline, and their institution’s readiness, allows it to be.

Subject of Research: The influence of artificial intelligence use on the academic learning behaviour of undergraduate students in Nigeria

Article Title: Artificial intelligence use and its influence on academic learning behaviour among undergraduate students in Nigeria

Article References: Chigozie, N. G. (2026). Artificial intelligence use and its influence on academic learning behaviour among undergraduate students in Nigeria. Discover Education, 5(1), Article 1007. https://doi.org/10.1007/s44217-026-02210-5

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02210-5

Keywords: artificial intelligence, higher education, undergraduate students, Nigeria, learning behaviour, ChatGPT, AI literacy, self-regulated learning, technology acceptance model, academic integrity, qualitative research, digital divide