EMPIRICAL EVALUATION OF AI CODE GENERATION TOOL OUTPUT QUALITY FOR COMMON PYTHON PROGRAMMING TASKS IN FEDERAL POLYTECHNIC, OFFA

ABSTRACT This project, titled Empirical Evaluation of AI Code Generation Tool Output Quality for Common Python Programming Tasks in Federal Polytechnic, Offa. This research scrutinizes the capability and dependability of Artificial Intelligence tools in the realm of automated Python programming. The analysis specifically contrasts two leading platforms-ChatGPT and GitHub Copilot-to gauge their success in generating code that is not only accurate and efficient but also legible and easy to maintain for both academic and professional applications. The study utilized an empirical experimental methodology, challenging both AI systems with ten standardized Python problems via natural language prompts. The resulting code was executed and rigorously evaluated against key metrics: success rate, processing speed, readability scores, and the maintainability index. The data indicated that ChatGPT delivered superior precision and clarity, achieving an average accuracy of 94.7%. In contrast, GitHub Copilot offered marginally faster execution speeds but lagged in code maintainability. The conclusions suggest that AI coding assistants are valuable educational aids that can streamline workflow and clarify algorithmic concepts for students and developers. However, the study cautions that human oversight remains essential to verify accuracy and avoid dependency. Ultimately, this paper argues for the inclusion of AI technologies, such as ChatGPT, within the Computer Engineering syllabus at Nigerian polytechnics. Such adoption is projected to modernize instructional methods, foster creativity, and better equip students for the demands of the contemporary digital workforce.