International Journal of Science, Engineering and TechnologyAuthors: Debasish Sarkar
Abstract: The rapid integration of Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) into academic inquiry has triggered an unprecedented paradigm shift in educational research. While AI-augmented literature synthesis, automated data coding, and language refinement offer remarkable productivity gains, they simultaneously introduce profound epistemic and methodological vulnerabilities. This doctoral dissertation provides an exhaustive, critical literature review exploring the complex interplay among Epistemic Trust, Methodological Opacity, and Hallucination Risks in modern educational scholarship. Synthesizing extensive peer-reviewed literature published between 2020 and 2024, this study adopts a virtue epistemology framework to analyze how algorithmic cognitive offloading impacts scholarly rigor. The findings reveal a growing systemic tension between productivity acceleration and epistemic vigilance: heavy reliance on generative tools correlates with a decline in primary-source verification, increased vulnerability to algorithmic hallucinations, and an over-reliance on opaque "black-box" data interpretations. Furthermore, this study evaluates current institutional policy responses across international higher education frameworks, highlighting a critical deficit in systematic verification protocols and doctoral AI literacy training. To resolve these systemic vulnerabilities, this review formulates the Epistemic Verification and Transparency Protocol (EVTP), providing actionable guidelines for researchers, journal editors, and university ethics committees to preserve scholarly integrity in an increasingly automated academic landscape
DOI: https://doi.org/10.5281/zenodo.23324179
