BOUCHEHIT , AymenBOUANINBA ,Nadine ; BOUGAMOUZA , Fateh2024-10-142024-10-142024http://dspace.univ-skikda.dz:4000/handle/123456789/2669Artificial Intelligence (AI) is revolutionizing numerous fields, including education, by enabling the creation of highly sophisticated and coherent texts through advanced language models. This technological advancement presents a dual-edged sword for educational institutions: while it offers new opportunities for learning and content creation, it also raises significant challenges regarding academic integrity. Students can now easily utilize AI tools to complete their assignments, thereby blurring the lines between genuine learning and academic dishonesty. This research aims to address the critical issue of detecting AI-generated texts to ensure the authenticity of academic work and the accurate assessment of students' knowledge and skills. By developing a model specifically designed to distinguish between human-written and AI-generated texts, we seek to uphold the integrity of academic evaluations. Our approach involves leveraging state-of-the-art AI models, such as AraBERT and transformers, to train the detection model. The study will encompass a comprehensive review of existing literature on AI content creation and detection, detailed analyses of the distinctive features of AI-generated texts, and the application of advanced techniques to identify these features. We will conduct experiments to test the effectiveness of our model and discuss the results obtained. Through this research, we aim to contribute valuable insights and practical solutions to the ongoing challenge of maintaining academic honesty in the age of AI.enMasters degree Thesis