Using Artificial Intelligence Tools in Scientific Research: A Systematic Review and Designing a Proposed Model.

Abstract

This study aimed to conduct a systematic review using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) criteria on the role of artificial intelligence tools in university scientific research from an international, Arab, and local perspective. This study was based on 251 articles extracted from leading academic databases, including Google Scholar, Dar Al- Mandumah, Crossref, ERIC, ScienceDirect, and SpringerOpen. The time frame for the selected studies and articles ranged from 2018 to 2024. The results of the analysis of the included studies showed that the most commonly used AI tools in scientific research are file and text search tools (64%) and machine translation (63%); while publishing and journal selection tools were not used at all, indicating a research gap in this area. The roles of AI tools focused on improving decision-making (84%), supporting data collection (48%) and idea generation (39%), with limited presence in research design and writing, reflecting a focus on the preparatory stages of research rather than the executive stages. The results also showed that the most prevalent ethical dimensions are fairness (73%), academic integrity (63%), privacy and security (57%) and transparency (53%), with little attention paid to specific practical dimensions such as avoiding plagiarism and harm, and an almost complete absence of the principle of privacy for researchers and examinees. The study also presented a proposed model for integrating AI tools into scientific research, based on the study's findings that identified the most important tools to use. The model combines research stages, appropriate tools, and academic and ethical roles. The study recommended the need for clear disclosure of the use of artificial intelligence tools at all stages of research, documentation of the traceability of analysis decisions, and emphasized the mandatory human review of any generated outputs and the avoidance of over-reliance on tools in sensitive interpretive decisions and plagiarism. It also recommended that universities in higher education adopt a multi-level governance model that measures the types and levels of AI tool use in scientific research, links them to research policies and publication ethics, and measures research performance quality with clear indicators.