Exploring Artificial Intelligence and Related Factors Among University Students

3. Data Description The dataset comprises responses from a cross-sectional online survey completed by 658 undergraduate and postgraduate students at Kwame Nkrumah University of Science and Technology (KNUST), Kumasi, Ghana, between June and July 2026. Participation was voluntary, and respondents were informed of the study’s purpose prior to completing the questionnaire. Demographic information covers gender, age group, college/discipline, mode of study, and current level of study. Of the participants, 354 (53.8 %) identified as male, 277 (42.1 %) as female, and 27 (4.1 %) preferred not to disclose gender. The majority (51.4 %) were aged 21–26 years. Students represented all six colleges of the university, with the largest proportion drawn from the College of Art and Built Environment (35.9 %). Most respondents (79.3 %) studied full-time on campus; 19.6 % were postgraduate students, while the remainder were distributed across undergraduate Years 1–4. Additional variables record exposure to formal AI training or education, frequency and duration of AI tool use, primary generative AI tools employed, main purposes of use, and preferred devices. Only 20.5 % of respondents reported having received formal AI training. Weekly (“often”) use was the most common pattern (44.1 %), followed by daily use (28.6 %). ChatGPT was the predominant tool (81.3 %), and the most frequent purposes were coursework (74.9 %) and studying (71.4 %). The majority of students accessed AI tools via both laptop/desktop and smartphone (65.3 %). AI literacy was measured across four constructs derived from multi-item scales: Awareness (AIA_MEAN), Use (AIU_MEAN), Evaluation (AIEV_MEAN), and Ethics (AIET_MEAN). Each construct was scored on a 1–7 scale. Mean scores ranged from 5.01 (Ethics) to 5.35 (Awareness), indicating moderately high levels overall. All pairwise Pearson correlations among the four constructs means were positive and statistically significant (p < .001), with coefficients ranging from 0.417 to 0.627. The deposited materials include the raw item-level responses, computed construct means, descriptive and correlational outputs, the full survey questionnaire, and a variable codebook. These resources permit independent re-scoring, alternative modelling approaches, and cross-study comparisons. The dataset is archived in the Mendeley Data Repository. Table 1 illustrates the socio-demographic profiles of the respondents who participated in the study and the frequencies(counts) of the participants of all variables.