More Reliable Literature Reviews

Although the previously mentioned AI systems can be helpful for conducting a literature search quickly, they, as already noted, have some limitations. Depending on the academic work, a more reliable and higher-quality literature review may be required.

As we have seen, AI systems rely heavily on open-access publications. Therefore, many relevant and influential articles that are not open access and are behind paywalls of academic publishers are not provided as results by AI research assistants. To access these academic articles, universities worldwide require subscriptions.

Students, professors, and researchers at Freie Universität Berlin, for example, can use the Primo Library Portal to search for and access books, journal articles, and conference papers. Alternatively/additionally, students, professors, and researchers can directly access the academic databases to which the university subscribes and conduct their literature search while connected to the campus network or via VPN.

Creating the Search Query

Regardless of whether the search is conducted via the Primo Library Portal or directly in academic databases, it is necessary to formulate a high-quality search query in order to increase the likelihood of finding relevant works. AI systems can be helpful here. A search query is a combination of keywords and Boolean operators entered into the search field of a library database. Researchers can use LLMs to support the process of creating such queries. See the following example:

Benmamoun, Mamoun. “Generative AI in International Business Research: A Guide to Ethical and Responsible Application.” Thunderbird International Business Review 67, no. 1 (2025): 139–46. https://doi.org/10.1002/tie.22415.

Boolean expressions are nothing more than multiple search terms combined with the operations “AND” or “OR.” Boolean expressions can be used either to broaden the search to obtain more results or to narrow it in order to retrieve only results that contain the desired information. In most search engines, this function can be found under “Advanced Search.” Primo at FU Berlin also offers this function, allowing users to easily adjust their search. In the following example, 634 results were found using the three specified keywords, whereas entering only the first two keywords yielded 4,296 results.

Selection Process

After conducting the search with AI research assistants such as Consensus or directly in academic databases such as Web of Science, researchers must select which articles they wish to analyze.

Both when using AI research assistants and when conducting direct searches in academic databases, many unrelated works are found. Researchers must therefore use inclusion and exclusion criteria to include articles in the literature review or exclude them (if they are not relevant).

Currently, there are AI systems that can assist researchers in accelerating the selection process and reducing their workload. Some examples of such AI systems are Abstrackr and Rayyan. Abstrackr helps save time by minimizing the risk of overlooking relevant articles and supports one or two reviewers. Rayyan, a widely used web-based software platform, integrates AI tools to facilitate the screening process. However, it is important to emphasize that although these AI systems demonstrate good accuracy, they may incorrectly classify an article as not relevant. Therefore, it is important that researchers review the decisions made by these AI systems.

In conclusion, for more reliable literature reviews, researchers increasingly use both AI-supported research assistants and academic databases. Some examples of studies that used mixed strategies for literature searches:

• Arcas, Vasile Calin, Anca Maria Fratila, Doru Florian Cornel Moga, Iulian Roman-Filip, Ana-Maria Cristina Arcas, Corina Roman-Filip, and Mihai Sava. “A Literature Review and Meta-Analysis on the Potential Use of miR-150 as a Novel Biomarker in the Detection and Progression of Multiple Sclerosis.” Journal of Personalized Medicine 14, No. 8 (August 2024): 815. https://doi.org/10.3390/jpm14080815.
• Rodríguez-Galán, Germán, and Jenny Torres. “Personal Data Filtering: A Systematic Literature Review Comparing the Effectiveness of XSS Attacks in Web Applications vs Cookie Stealing.” Annals of Telecommunications 79, No. 11–12 (December 2024): 763–802. https://doi.org/10.1007/s12243-024-01022-8.
• Jiménez, Amalio, Frederick R. Carrick, Norman Hoffman, and Monèm Jemni. “The Impact of Low-Level Laser Therapy on Spasticity in Children with Spastic Cerebral Palsy: A Systematic Review.” Brain Sciences 14, No. 12 (December 2024): 1179. https://doi.org/10.3390/brainsci14121179.
• Kádár, Béla, and Erika Jáki. “COVID-19 and SMEs: An Umbrella Review of Systematic Literature (2020–2024) and Future Directions for Entrepreneurship: Introduction to the ‘The Entrepreneurial Landscape in the Post-COVID Era: Insights, Challenges, and Future Perspectives’ Special Issue.” Society and Economy 46, No. 4 (7 December 2024): 323–41. https://doi.org/10.1556/204.2024.00017.