Fully AI-Generated Research: Testing the Limits of Autonomous Paper Writing

Abstract

Waseem Khoso

The rapid advancement of large language models (LLMs) has ushered in a new era of automated academic writing, challenging traditional notions of authorship and scholarly production. This study presents a fully AI-generated research paper as a case study to critically evaluate the current capabilities and limitations of artificial intelligence in academic research. Using a multi-phase prompting methodology with GPT-4, we demonstrate that contemporary LLMs can autonomously produce manuscripts that meet basic structural and stylistic requirements of academic writing, including coherent argumentation, literature synthesis, and formal citation formatting. However, our analysis reveals significant limitations in the areas of original insight generation, factual accuracy, and ethical citation practices. The paper highlights three critical tensions: (1) between productivity gains and academic integrity, (2) between linguistic fluency and substantive depth, and (3) between automation potential and the need for human oversight. We identify key risks including citation hallucinations, undisclosed automation, and the erosion of critical thinking skills. The study concludes with a proposed framework for responsible AI use in academia, recommending tiered implementation guidelines, enhanced disclosure requirements, and hybrid human-AI collaboration models.

These findings contribute to ongoing discussions about research ethics in the age of generative AI and provide practical recommendations for researchers, institutions, and publishers navigating this transformative period in scholarly communication.

PDF