Dataset on Large Language Models and GenAI Adoption Among Nigerian Teachers

Our dataset provides empirical insights into the adoption of Large Language Models (LLMs) and Generative AI (GenAI) in education, specifically examining Nigerian in-service teachers’ behavioural intentions (BI) to integrate ChatGPT into their instructional practices. The study adopts a hybrid analytical approach, combining Partial Least Squares Structural Equation Modelling (PLS-SEM) and Artificial Neural Networks (ANN) to explore both linear and nonlinear relationships among key adoption predictors. Guided by the Technology Acceptance Model (TAM), our research hypothesises that Perceived Ease of Use (PEU) and Perceived Usefulness (PUC) positively influence teachers' behavioural intentions, while Technology Anxiety (TA) and Privacy Issues (PIU) negatively impact adoption. Additionally, we investigate the role of Attitude Toward ChatGPT (ATC), Teachers’ Trust in ChatGPT (TTC), and peer influence (YCC) in shaping adoption behaviour.

Our findings from PLS-SEM indicate that PUC emerged as the strongest predictor of behavioural intention, suggesting that teachers are more likely to adopt ChatGPT if they perceive it as beneficial to their teaching. Conversely, TA significantly negatively influenced BI, highlighting that anxiety related to AI technology acts as a major barrier to adoption. Additionally, peer influence (YCC) was found to positively impact BI, reinforcing the importance of social support in facilitating technology adoption among educators. Although PEU had a weaker negative effect on BI in the PLS-SEM model, ANN results revealed a more complex picture, ranking PEU as the most important predictor, followed by ATC and PUC. This suggests that while usability concerns may appear less significant in a linear framework, they become crucial when considering nonlinear interactions.

The dataset can be used for multiple research applications, including AI adoption studies, predictive modelling using ANN, and comparative analyses across different educational contexts. Researchers can utilise it to examine the direct and indirect relationships influencing AI adoption, while policymakers and education stakeholders can derive actionable insights to design AI training programmes that emphasise usability and trust while addressing technology-related anxieties. Furthermore, this dataset provides a foundation for future cross-cultural studies on AI integration in resource-constrained educational environments, offering a benchmark for understanding teacher adoption behaviour and the broader implications of AI in education.