Prevalence and Factors Influencing ChatGPT Use During the Research Process Among Undergraduate Health Students in Tanzania: A Cross-Sectional Study
Ipyana Mbaga Kajembula, Kornel Izidory Matheo, Peter Damian Simchimba, Diana Damasi Massawe, Aminieli Itaeli Usiri, Joel Seme AmbikileBackground
ChatGPT is a large language model developed by OpenAI and launched November 30, 2022, with potential to transform healthcare education and research. However, integration into academic settings raises concerns about integrity and ethics. This study assessed prevalence and factors influencing ChatGPT use during the research process among undergraduate health students in Tanzania.
Methodology
A cross-sectional study was conducted among undergraduate health students in Tanzania mainland universities. Multi-stage sampling recruited participants who completed an online structured questionnaire. Data were collected using the validated Technology Acceptance Model Edited to assess ChatGPT Adoption (TAME-ChatGPT). Descriptive statistics summarized characteristics, while Independent sample t-tests and one-way ANOVA compared mean attitude scores. Poisson regression with robust standard errors assessed factors associated with ChatGPT use, with significance at p < 0.05.
Results
Of the 422 students invited, 400 (95%) responded; 247 (61.8%) were male, with a mean age of 24.82 ± 2.13 years. The prevalence of ChatGPT use during the research process was 63.0%, with students primarily using it for idea generation (57.1%), information summarization (52.8%), and writing assistance (29.4%). Among users, perceived ease of use (4.01 ± 0.79) and perceived usefulness (3.56 ± 0.71) recorded the highest TAME-ChatGPT construct scores. In multivariable Poisson regression, ChatGPT use was significantly more likely among students from private universities (aRR = 1.23, 95% CI: 1.057–1.433, p = 0.008), those at institutions acknowledging ChatGPT use (aRR = 1.34, 95% CI: 1.094–1.646, p = 0.005), and those unsure of their institution’s stance (aRR = 1.34, 95% CI: 1.051–1.709, p = 0.018), compared to students at non-acknowledging institutions. Students with more than five peers per supervisor were also more likely to use ChatGPT (aRR = 1.20, 95% CI: 1.029–1.404, p = 0.020).
Conclusion
ChatGPT use is prevalent among Tanzanian undergraduate health students during the research process, driven primarily by institutional context and supervisory dynamics. Institutions and national education authorities should establish clear, evidence-based guidelines to ensure the ethical and effective integration of AI tools in academic research.