Artificial Intelligence and the Paper Writing Revolution: A Literature Analysis of the IMRAD Model in the Age of Large Language Models

Abstract

The academic landscape is undergoing a radical transformation due to the infiltration of artificial intelligence (AI) and large language models (LLMs), which are reshaping the mechanisms of scientific knowledge production, from literature exploration to manuscript writing. This article presents a specialized, systematic literature analysis examining the impact of this technological revolution on the structure of the paper according to the IMRAD model (Introduction, Methodology, Results, Discussion). Based on a meticulous review of more than 40 recent studies (2024-2026), the research presents an analytical framework divided into four main axes: (1) Intelligent tools for literature discovery and citation analysis, and their classification in detailed analytical tables; (2) Applications of LLMs in data analysis and extraction, with a focus on critiquing the phenomenon of academic "hallucination"; (3) The transformation of manuscript writing and review with the aid of intelligent agents (Multi-Agent Frameworks); (4) Ethical Dilemmas and Flaws in Artificial Content Detectors. The results reveal that large linguistic models demonstrate high efficiency in structured tasks (such as title and abstract scanning, where recall rates reach 100%, reducing workload by up to 97.3%). However, they remain fragile and unreliable in tasks requiring deep scientific judgment (such as extracting complex data, with recall rates ranging from 45.5% to 94.4%). The article concludes that responsible integration between humans and artificial intelligence, where these tools act as assistants to enhance capabilities and not as replacements for human expertise, is the only way to ensure the integrity of scientific research in the foreseeable future.