Date of Award
10-2025
Culminating Project Type
Starred Paper
Styleguide
apa
Degree Name
Computer Science: M.S.
Department
Computer Science and Information Technology
College
School of Science and Engineering
First Advisor
Maninder Singh
Second Advisor
Mark Gill
Third Advisor
Andrew Anda
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Abstract
The research investigates whether artificial intelligence systems perform systematic literature reviews (SLRs) with the same level of quality as human researchers. The same question, with the same scope and eligibility requirements, was applied to both a conventional manual approach and an AI assisted approach using a comparative experimental design. 29 studies total eight from manual screening and 21 from AI assisted methods were included in the two approaches. The quantitative results demonstrate AI achieved a 2.08× speed-up because it improved precision to 70% from 27% and increased screening to full text yield to 70% from 52% and cut the total review time in half from 25 hours to 11 hours. The results showed that manual studies generated summaries with 0.95 fidelity while achieving 4.7 ± 0.5 relevance scores which surpassed 0.82 and 3.5 ± 0.6 respectively. Error analysis results showed that AI output contained two primary error types which included omissions that happened in 25% of cases and misinterpretations that appeared in 10% of instances. The study showed that human verification served as a vital process to verify AI output accuracy. The thematic synthesis produced four main results which demonstrated that AI technology accelerates research discovery yet reduces the accuracy of selection processes and different tools produce different results about reproducibility and quality control stands as a fundamental requirement and the most effective outcomes stem from combining human intelligence with artificial intelligence workflow systems. The study's overall conclusion is that AI can improve the speed and scope of SLRs, but only if it is integrated into open, researcher in-the-loop procedures.
Recommended Citation
Niraula, Nikesh, "Reimagining Systematic Literature Reviews: A Comparative Study of AI-Assisted and Manual Approaches" (2025). Culminating Projects in Computer Science and Information Technology. 67.
https://repository.stcloudstate.edu/csit_etds/67
Comments/Acknowledgements
I want to express my deepest appreciation to my supervisor who provided ongoing guidance and support and valuable feedback during the entire research period. Their professional guidance together with their motivational support played a crucial role in determining both the direction and excellence of this thesis.
I appreciate the academic support from St. Cloud State University which gave me access to all necessary resources and facilities to complete this research project