The integration of artificial intelligence (AI) into educational frameworks has emerged as a transformative force in contemporary pedagogy, particularly in developing nations where traditional educational systems face significant resource constraints. In Nigeria, the National Information Technology Development Agency’s release of the National Artificial Intelligence Strategy (NAIS) in 2024 explicitly recognized education as a priority area for AI application, emphasizing evidence-based approaches to technology integration (Adeoye & Babatunde, 2024). This strategic framework has catalyzed increased attention to understanding how AI tools influence academic outcomes in Nigerian educational institutions.

Within the Nigerian educational context, teacher education institutions represent a critical junction where AI literacy and competency development can have multiplier effects. The Nigeria Certificate in Education (NCE) program, which prepares primary and secondary school teachers, serves approximately 8,500 students at Sa’adatu Rimi College of Education in Kano State. Recent empirical evidence from Nigerian educational contexts has demonstrated remarkable success in AI implementation, with pilot programs utilizing generative AI to support learning yielding improvements equivalent to nearly two years of typical schooling achieved in just six weeks (World Bank, 2025). Nigerian students participating in these programs have expressed enthusiastic support for AI integration, with participants noting that “AI helps us to learn, it can serve as a tutor, it can be anything you want it to be, depending on the prompt you write” (World Bank, 2025).

Despite growing global recognition of AI’s transformative potential and increasing availability of AI tools to Nigerian students, a significant knowledge gap exists regarding specific impacts on academic performance among NCE students. While international research demonstrates that AI technologies can significantly enhance learning outcomes when properly integrated (Huang et al., 2024), the specific mechanisms and contextual factors determining success remain understudied in African educational contexts. This gap is particularly concerning given that a considerable number of educational institutions in Nigeria have yet to leverage AI technologies effectively (Okonkwo & Ade-Ibijola, 2024).

The research problem extends beyond simple adoption metrics to encompass the complex interplay between AI usage patterns, academic performance outcomes, and contextual factors unique to Nigerian teacher education. Infrastructure limitations, digital literacy disparities, and cultural factors represent potential moderators of AI’s educational impact in Nigerian settings (Adebayo & Ogundimu, 2024). Furthermore, the digital divide documented by researchers shows substantial variations in internet access and device availability among students in Northern Nigerian educational institutions, suggesting that AI’s benefits may not be equally accessible to all students (Bello & Garba, 2024).

This study addresses these gaps by examining the impacts of artificial intelligence on academic performance of NCE students at Sa’adatu Rimi College of Education, Kano, with specific attention to usage patterns, performance relationships, perceived benefits, challenges, and demographic moderation effects. The research contributes to both theoretical understanding and practical applications of AI in developing country educational contexts while informing evidence-based policy development for Nigerian teacher education institutions.

Literature Review

This investigation adopts an integrated theoretical framework combining elements from the Technology Acceptance Model (TAM), Constructivist Learning Theory, and Cognitive Load Theory to understand AI’s impacts on student academic performance. The Technology Acceptance Model serves as the foundation for understanding how and why NCE students adopt AI tools, with perceived usefulness and perceived ease of use as core constructs explaining adoption patterns and usage intensity (Davis & Johnson, 2024). Recent adaptations of TAM specifically for AI educational contexts have incorporated additional factors such as trust in AI systems, concerns about privacy and academic integrity, and perceived impact on learning autonomy (Brown et al., 2025).

Constructivist Learning Theory provides the pedagogical lens for evaluating AI’s impact on learning processes, emphasizing that learning occurs through active construction of knowledge rather than passive reception of information. Research applying constructivist principles to AI-enhanced learning has shown that when AI tools are designed to promote active engagement and reflection rather than passive consumption, they can significantly enhance learning outcomes (Vygotsky Institute, 2024). Cognitive Load Theory offers insights into cognitive mechanisms through which AI tools influence academic performance by distinguishing between intrinsic cognitive load related to learning content, extraneous cognitive load related to instructional design, and germane cognitive load related to meaningful learning processes (Sweller & Clark, 2024).

Artificial Intelligence in Education (AIED) represents a multidisciplinary field combining computer science, cognitive psychology, and educational theory to create intelligent systems supporting learning and teaching processes. AI in education has evolved into a substantial body of literature with diverse perspectives, encompassing various AI technologies including machine learning, natural language processing, computer vision, and intelligent tutoring systems designed to enhance educational experiences (García-Peñalvo et al., 2024). UNESCO recognizes that artificial intelligence has the potential to address some of the biggest challenges in education today, innovate teaching and learning practices, and accelerate progress towards Sustainable Development Goal 4 (UNESCO, 2025).

Academic performance, as a multifaceted construct, encompasses various indicators of student learning outcomes including grades, test scores, assignment quality, critical thinking abilities, and overall educational achievement. In the context of Nigerian teacher education, academic performance extends beyond traditional metrics to include professional competencies, pedagogical skills, and readiness for classroom practice. The measurement of academic performance in the digital age has become increasingly complex as educators grapple with distinguishing between authentic student work and AI-assisted outputs (Chaudhary et al., 2025).

Methodology

This study employed a quantitative research design utilizing a descriptive survey approach to examine AI’s impacts on NCE student academic performance. Grounded in the positivist research paradigm, the investigation emphasizes objective measurement, empirical observation, and statistical analysis to understand social phenomena. The cross-sectional survey design collected data at a single point in time to capture current AI usage patterns and their relationship with academic performance indicators, aligning with recent methodological trends in AI education research where such surveys have proven effective for understanding technology adoption and impact patterns (HEPI, 2025).

The target population comprised all NCE students enrolled at Sa’adatu Rimi College of Education, Kano, during the 2024/2025 academic session, totaling approximately 8,500 students distributed across three levels (NCE I, II, and III) and various subject specializations including Arts, Sciences, Social Sciences, Technical Education, and Vocational Education. Stratified random sampling ensured representative selection across key demographic and academic variables, with primary stratification based on academic level, subject specialization, and gender.

Sample size was calculated using the Yamane formula for finite populations: n = N / (1 + N(e)²), where N = 8,500 and e = 0.05, yielding n = 382. To account for potential non-response and ensure adequate representation across strata, the sample size was adjusted upward by 20%, resulting in a final target sample of 458 students. This sample size aligns with recent AI education surveys that have successfully employed similar approaches, with the HEPI study utilizing 1,041 undergraduate students to achieve reliable results (HEPI, 2025).

Primary data was collected through a structured questionnaire administered both online via Google Forms and offline through paper-based formats to accommodate varying levels of internet access and technological familiarity among participants. The questionnaire design followed best practices established in recent AI education research, incorporating validated scales and instruments where possible while adapting content to the specific Nigerian context. Data analysis employed SPSS version 29.0 for comprehensive statistical analysis including descriptive statistics, reliability testing (Cronbach’s alpha), correlation analysis (Pearson’s r), multiple regression, and moderation analysis. All statistical tests were conducted at a 95% confidence level (α = 0.05).

The statistical model specification included academic performance as the dependent variable measured through multiple indicators including cumulative grade point average (CGPA), subject-specific performance scores, assignment quality ratings, and research project completion rates. Independent variables comprised AI usage frequency, AI tool variety, AI usage purpose, and AI self-efficacy. Control variables included digital literacy level, academic level, gender, age, subject specialization, and socio-economic status proxies.

Results and Findings

The survey achieved an exceptional response rate of 89.96%, with 412 valid responses received from 458 distributed questionnaires. This response rate compares favorably with recent AI education surveys, including the HEPI Student Generative AI Survey 2025 which achieved 84% among UK undergraduate students (HEPI, 2025). Response rates were remarkably consistent across academic levels, ranging from 88.82% to 90.85%, and across subject specializations, ranging from 89.47% to 90.43%, indicating strong engagement and minimal non-response bias.

Table 1: Demographic Profile of Respondents (N = 412)

The demographic profile reveals a well-balanced sample with slight female majority (51.9%), closely mirroring typical composition of Nigerian teacher education institutions. Nearly 60% of students fall within the traditional 18-22 age range, with substantial representation of older students (40.5% aged 23+) reflecting delayed entry patterns common in Nigerian education. Equal distribution across academic levels and subject areas confirms successful stratified sampling and enhances generalizability of findings.

Table 2: AI Tool Usage and Academic Performance

Analysis reveals that 84.95% (N = 350) of NCE students actively use AI tools for academic purposes, representing remarkably high adoption rates exceeding many international benchmarks. ChatGPT emerges as the dominant platform with 66.1% usage rate, while usage intensity shows 23.1% daily users and 61.9% weekly or monthly users, indicating purposeful rather than compulsive engagement. Academic performance indicators demonstrate healthy distribution with current CGPA mean of 3.12 showing above-average achievement and slight improvement from previous semester (3.08).

Table 3: Purposes of AI Usage in Academic Activities (N = 350)

Research and information gathering dominates AI usage (82.0%), positioning AI primarily as a knowledge discovery tool rather than content creation substitute. Assignment writing assistance ranks second (66.9%), while language translation (56.6%) likely reflects multilingual demands of Nigerian education where students navigate between English academic requirements and native languages. The hierarchy of usage reveals immediate practical needs prioritized over developmental activities.

Table 4: Correlation Analysis – AI Usage and Academic Performance

Note: Correlation is significant at the 0.01 level (2-tailed)

All correlations are statistically significant at the 0.01 level, indicating robust relationships between AI engagement and academic success. AI self-efficacy emerges as the strongest predictor across all performance indicators (r = 0.398 to 0.467), aligning with self-efficacy theory and suggesting that confidence and competence in AI usage translate directly into academic benefits. Usage frequency shows consistent moderate correlations (r = 0.284 to 0.378), with the strongest relationship to research projects (r = 0.378), indicating AI tools are particularly valuable for complex, inquiry-based tasks requiring information synthesis and analysis.

Table 5: Multiple Regression Results – Predictors of Academic Performance

Model Summary: R² = 0.387, Adjusted R² = 0.379, F = 51.23, p < 0.001

The multiple regression model explains 38.7% of variance in academic performance, representing substantial explanatory power for educational research. AI self-efficacy emerges as the strongest predictor (β = 0.298, p < 0.001), indicating a one-standard-deviation increase in AI self-efficacy associates with 0.298 standard-deviation increase in academic performance. The significant but smaller coefficients for AI usage frequency (β = 0.189) and digital literacy (β = 0.156) indicate these factors contribute independently but are less influential than AI self-efficacy. Academic level and gender represent significant control variables requiring consideration in AI education research.

Table 6: Perceived Benefits and Challenges of AI Usage

Note: Measured on 5-point Likert scale (1 = Strongly Disagree, 5 = Strongly Agree)

The ranking of perceived benefits reveals pragmatic student perspectives with faster information access (M = 4.23) and improved research capabilities (M = 4.12) addressing fundamental challenges in developing country educational contexts where traditional resources may be limited. Students recognize AI’s particular value for complex information tasks requiring synthesis and analysis. Primary challenges include inconsistent internet connectivity (M = 4.45) reflecting Nigeria’s digital infrastructure issues, and academic integrity concerns (M = 3.87) demonstrating encouraging ethical awareness requiring institutional guidance.

Table 7: Moderation Effects on AI Usage-Performance Relationship

Digital literacy emerges as the most significant moderator (β = 0.234, p < 0.001), indicating students with stronger technological skills experience significantly greater academic benefits from AI usage. Gender moderation (β = 0.124, p = 0.020) suggests female students may derive greater academic benefits, possibly reflecting different usage strategies or learning preferences. Academic level moderation (β = 0.156, p = 0.003) confirms AI’s benefits compound with advancement as more experienced students leverage tools more effectively through developed critical thinking skills and deeper subject knowledge.

Discussion

The finding that 84.95% of NCE students have used AI tools for academic purposes represents remarkably high adoption rates exceeding many international benchmarks and challenges assumptions about technology adoption in developing country educational contexts. This aligns with recent global trends showing rapid AI adoption among university students, where studies report usage rates ranging from 70% to 89% across different educational contexts (HEPI, 2025; Sun & Zhou, 2024). The dominance of ChatGPT as the preferred AI tool reflects global patterns where conversational AI tools have achieved mainstream adoption.

The study provides empirical evidence that AI tools, when used appropriately, have positive impacts on NCE student academic performance. The moderate to strong positive correlations (r = 0.284 to 0.467) demonstrate AI serves as valuable educational resource rather than detrimental dependency, aligning with meta-analytic findings showing consistent positive effects across diverse educational contexts (Kim & Lee, 2025). The strongest correlation between AI self-efficacy and academic performance (r = 0.467) aligns with Technology Acceptance Model predictions and recent empirical findings showing confident and skilled AI usage leads to better outcomes (Chen et al., 2025).

The particularly strong correlation between AI usage and research project scores (r = 0.378) is noteworthy for teacher education contexts, as research skills are crucial for evidence-based teaching practices. This finding suggests AI tools may be particularly valuable for developing analytical and inquiry skills needed for effective teaching. The multiple regression model explaining 38.7% of variance represents substantial explanatory power, with AI self-efficacy as strongest predictor (β = 0.298) reinforcing that quality of AI interaction matters more than mere usage frequency.

The emphasis on practical benefits such as faster information access and improved research capabilities addresses fundamental challenges in Nigerian educational contexts where library resources and internet connectivity may be limited. However, the moderate rating for enhanced creativity (M = 3.23) suggests students may not yet fully recognize or utilize AI’s potential for creative academic work, providing opportunities for educational interventions exploring AI’s creative applications while maintaining academic integrity.

Infrastructure and ethical challenges reveal continued digital divide affecting Nigerian educational institutions, with inconsistent internet connectivity as primary barrier (M = 4.45) emphasizing need for infrastructure development. Academic integrity concerns ranking second (M = 3.87) demonstrate encouraging ethical awareness creating opportunities for developing responsible usage practices. The relatively low concern about privacy and data security (M = 2.98) may indicate limited awareness requiring enhanced digital citizenship education.

The significant moderation effects for digital literacy, academic level, and gender indicate differential impacts across student groups requiring inclusive AI education strategies addressing varying levels of technological access. Female students experiencing greater benefits challenges assumptions about gender and technology, suggesting AI tools may be particularly well-suited to collaborative and research-oriented learning approaches. The universality of benefits across age and subject specialization supports broad-based AI integration initiatives in teacher education programs.