Biography
Dr. Mehdi Kaddouri is a professor at Mohammed First University, Morocco, specializing in educational technology and digital learning. He earned his PhD from Paul Valéry University of Montpellier (France). Currently, he serves as the Director of the CEDUC-URL-CNRST-19 Laboratory and heads the E-learning Center at Mohammed First University. His research focuses on the integration of technology in education, digital practices in learning environments, and educational semiotics. Dr. Kaddouri examines educational technology implementation, digital learning processes, and semiotic approaches in education, contributing to the understanding of digital transformation in educational settings and its impact on teaching and learning practices in higher education.
Conference contributions 1
WHEN AI OVERCORRECTS:
A theoretical examination of learner self-efficacy in academic writing
The growing integration of AI-powered writing tools in higher education has introduced new and underexplored challenges for academic writing development. While these tools are considered as valuable feedback resources capable of enhancing students' writing quality, this presentation argues that their systematic tendency to suggest modifications to already correct and academically appropriate writing constitutes a form of overcorrection with serious pedagogical consequences. Unlike traditional corrective feedback, which targets genuine errors, AI tools such as ChatGPT and Gemini consistently propose alternative vocabulary, restructured sentences, and stylistic changes regardless of whether the original writing meets academic standards. This pattern of unnecessary intervention risks sending students a misleading message — that their language choices are perpetually insufficient. Drawing on Bandura's (1997) self-efficacy theory, Truscott's (1996) critique of written error correction, Hyland's (2002) academic voice framework, and Holec's (1981) learner autonomy theory, this presentation theorizes that persistent AI suggestion creates a false impression of inadequacy in students, which leads them to distrust their own linguistic judgments and undermine their confidence as academic writers. Constant AI intervention further erodes learner autonomy and silences the development of an authentic scholarly voice, carrying direct negative consequences for long-term academic writing performance. This presentation calls for raising critical awareness among teachers and students regarding the overcorrective nature of AI feedback tools, advocating for a balanced and reflective approach that encourages students to trust their own competence rather than developing an uncritical over-reliance on AI-generated suggestions.