Establishing Ethical Guidelines for AI Usage in EFL Academic Writing

Abstract

This qualitative study investigates the ethical complexities of AI authorship among 25 EFL college learners, focusing on the tension between linguistic enhancement and the preservation of authorial voice. By mapping an "Authorial Spectrum," the research distinguishes between AI acting as a "Cradle" (technical scaffolding for grammar and syntax) versus a "Creator" (the automated generation of original conceptual ideas). Findings reveal a pervasive "Shortcut Trap," where deadline pressures drove 68% of participants toward intellectual dependency, allowing AI-generated structures to supersede their personal logical flow. However, the data also highlights the efficacy of "Socratic Prompting"—defined as iterative, question-based AI interaction—in safeguarding unique cultural nuances from "algorithmic sanitization" (the loss of identity through standardized AI output). This method resulted in a 40% higher retention of original idiomatic expressions compared to direct, one-off prompting. Consequently, the study advocates for a process-oriented disclosure model using transparency logs to address "digital ghostwriting." By moving beyond a prohibition-based mindset, the research positions AI as a tool to sharpen the human filter rather than replace the learner’s identity. Ultimately, these findings ensure that the diverse voices of EFL learners remain the primary drivers of academic discourse in an increasingly automated age.