HumanVsGenAI: AI Code Detection Performance in Educational Programming Tasks using DetectCodeGPT

The widespread adoption of AI coding tools in educational settings has created new challenges in assessing student work and maintaining academic integrity. This study aims to evaluate the effectiveness of AI code detection systems, specifically DetectCodeGPT, in distinguishing between AI-generated and human-written solutions in programming assignments. Through systematic analysis of detection performance across different types of programming problems, we seek to understand patterns that can inform better assignment design. Our methodology combines collection of paired AI and human solutions with rigorous detection analysis. The expected outcomes include guidelines for designing AI-resistant programming assignments and insights into detection system effectiveness. The project follows a structured 9-week plan, encompassing data collection, analysis, and documentation phases.