LARGE LANGUAGE MODELS IN SCIENTIFIC RESEARCH WORKFLOWS: A SYSTEMATIC LITERATURE REVIEW OF APPLICATIONS, CHALLENGES, AND FUTURE DIRECTIONS
Journal of Intelligent Systems and Applied Data Science · 06.07.2026 · 4 (1)
DOI: 10.64680/jisads.v4i1.65 ↗Kurz gefasst
Systematische Übersicht in der Zeitschrift Journal of Intelligent Systems and Applied Data Science, online erschienen am 06.07.2026. Verfasst von kolawole Olulana und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background
Large language models (LLMs) are increasingly being used in scientific research workflows, including literature review, hypothesis formulation, manuscript generation, and peer review. Notwithstanding their swift adoption, concerns exist over epistemic reliability, transparency, assessment, and the integrity of authorship.
Objectives
This review aims to collect, summarize systematic data on research process of integration of LLMs into scientific workstreams, evaluate their outputs, identify recurring methodological, ethical and epistemic problems encountered during their adoption and deployment.
Methods
To achieve this, a systematic literature review was performed following the guidelines of PRISMA 2020. An extensive search of the Scopus database revealed 36 peer-reviewed studies from 2018 to 2025. Analysing the included studies, which were coded using a structured coding framework that highlighted application stage, research domain, evaluation strategy, reported benefits and ethical and epistemic issues. The results were integrated into the synthesis process by thematic mapping and cross-study comparison.
Results
The results show that the majority of LLM adoption is currently focused mostly on text-oriented activities which mostly includes literature synthesis and scientific writing, whereas they have more limited application in hypothesis development, experimental design, and peer review methods. Evaluation practices differ significantly across studies and are often not standardized across studies, negatively affecting comparability and reproducibility. Common issues are hallucinated content, unverifiable citations, opaque behaviour from the model, and authorship and accountability ambiguity.
Conclusions
This review shows that LLMs have significant potential for lowering the cognitive burden to increase efficiency of scientific research workflow and to increase efficiency, but their implementation is epistemically fragile. To the above limitations, the review presents the LLM–Research Workflow Integration Framework (LRWIF), conceptualizing LLM use at stages of knowledge discovery, study design, scientific writing, and evaluation. The results underscore the importance of common evaluation baselines, transparent and repeatable AI pipelines, and stronger ethical governance with the objective of making LLMs a contributing factor not a detractor to scientific inquiry.
Zitieren
APA
Olulana, K., Pius, O. A., Du, C., & Abe, B. T. (2026). LARGE LANGUAGE MODELS IN SCIENTIFIC RESEARCH WORKFLOWS: A SYSTEMATIC LITERATURE REVIEW OF APPLICATIONS, CHALLENGES, AND FUTURE DIRECTIONS. Journal of Intelligent Systems and Applied Data Science, 4(1). https://doi.org/10.64680/jisads.v4i1.65
Vancouver
Olulana K, Pius OA, Du C, Abe BT. LARGE LANGUAGE MODELS IN SCIENTIFIC RESEARCH WORKFLOWS: A SYSTEMATIC LITERATURE REVIEW OF APPLICATIONS, CHALLENGES, AND FUTURE DIRECTIONS. Journal of Intelligent Systems and Applied Data Science. 2026;4(1). doi:10.64680/jisads.v4i1.65
BibTeX
@article{Olulana2026,
author = {Olulana, kolawole and Pius, Owolawi Afolabi and Du, Chunling and Abe, Bolanle Tolulope},
title = {LARGE LANGUAGE MODELS IN SCIENTIFIC RESEARCH WORKFLOWS: A SYSTEMATIC LITERATURE REVIEW OF APPLICATIONS, CHALLENGES, AND FUTURE DIRECTIONS},
journal = {Journal of Intelligent Systems and Applied Data Science},
year = {2026},
volume = {4},
number = {1},
doi = {10.64680/jisads.v4i1.65}
}
EUVIMED