Artificial Intelligence in Academic Writing: Opportunities and Risks from Planning to Publication

Abstract

With the recent prevalence and enhanced accessibility of Artificial Intelligence (AI) tools, the aim of this review is to assess both the benefits and risks of using AI tools in academic writing, thus allowing authors to make more informed decisions with regard to AI usage based on the current literature and the best available evidence.

Between February and April 2024, the authors conducted a narrative review of the academic literature using AI-focused keyword searches of databases including PubMed. Risks and benefits of using AI in academic writing were identified and subcategorized into four stages: planning, execution of research, drafting of a manuscript, and publication.

The literature suggests that AI tools, particularly large language models, provide several potential benefits at each stage of academic writing. This includes assistance in idea generation, data analysis, peer review, and drafting, with the potential to significantly improve overall efficiency. Significant challenges were also identified, including bias, plagiarism risk, and misleading AI-generated content (often referred to as hallucinations).

In conclusion, AI tools appear to present promising opportunities for improving academic writing and could potentially revolutionise the process in which academic research is conducted. Careful consideration of their limitations with legal and ethical implications is paramount—thus, the authors recommend that a collaborative effort led by the academic community is needed to establish best practice guidelines and regulatory frameworks for the responsible and effective implementation of AI tools in the process of scientific publications.