Artificial intelligence and evidence synthesis
Artificial intelligence (AI) is reshaping evidence synthesis literature reviews by impacting efficiency, consistency and scalability.
Rather than replacing researcher expertise, AI tools can augment traditional methods performed in a literature review such as preliminary searching and screening.
AI tools can support researchers in managing the increasing volume of studies while maintaining rigorous methodological standards.
Caution
The use of AI at any stage of a literature review must be reported to ensure transparency. It should be acknowledged that complete transparency may not be achievable with certain AI tools, and their application in reviews should be evaluated accordingly.
Additionally, it is crucial to adhere to both review methodological and reporting guidelines to maintain compliance.
Good practices in using AI tools
Responsible use of AI tools in a literature review is essential. Researchers should ensure they apply the following approaches when using AI tools:
- Critical engagement: Researchers should evaluate AI-generated recommendations instead of accepting them without question.
- Ethical considerations: AI-assisted research should align with Australian research codes, emphasising integrity, fairness and responsible AI deployment.
- Transparency: Use of AI tools should be documented in research methodologies to ensure reproducibility.
AI and Deakin research
Deakin University supports the responsible integration of AI in research through endorsed tools, library guidance and research integrity initiatives. The Deakin Library provides AI literacy resources, training and expert consultations to help researchers navigate AI in evidence synthesis effectively.
For further information on AI in research, check out our Generative AI: responsible use in research guide.
AI tools and literature reviews
There is a growth of AI tools that can be used in literature reviews, such as AI search engines, AI literature mapping and in screening tools. Increasingly, traditional databases are also incorporating AI search functionality.
AI search engines
AI search engines work with natural language or a conversational style search rather than traditional keywords. They can assist in identifying key articles during preliminary searching. These key articles can help you create a comprehensive search strategy to locate and collect all relevant material on your review topic from relevant information sources, such as key discipline databases.
Click on the plus (+) icons below to learn about some AI search engines.
Semantic Scholar
Semantic Scholar functions as a literature discovery, summarisation and current awareness tool.
Semantic Scholar is free.
Semantic Scholar sources data from Semantic Scholar Open Research Corpus (built from Publisher Partners, web crawls and other data providers).
Elicit
Elicit functions as a literature discovery, summarisation and data extraction tool. It can also generate research reports and automate steps in a systematic review.
Elicit is a freemium service, it's free to sign up. However, access to additional features comes at a cost to users.
Elicit sources data from Semantic Scholar.
Consensus
Consensus is an academic AI search engine. It provides summary overviews of literature and includes a consensus meter that indicates the level of agreement among papers on the search query.
Consensus is a freemium service, it's free to sign up. However, access to additional features comes at a cost to users.
Consensus sources data from Semantic Scholar.
Perplexity
Perplexity is an AI search engine, it summarises answers to search queries and provides links to sources. Deep Research functionality.
Perplexity is a freemium service, it's free to sign up. However, access to additional features comes at a cost to users.
Perplexity sources data by crawling the web.
Felo
Felo is a multi-lingual AI search engine, it summarises answers to search queries and provides links to sources. Deep Research functionality.
Felo is a freemium service, it's free to sign up. However, access to additional features comes at a cost to users.
Felo sources data by crawling the web.
AI literature mapping tools
AI literature mapping tools identify connections between papers, through citations, references, or through algorithms which surface papers similar to starting papers.
These tools serve a similar purpose to citation databases, such as Scopus or Web of Science, in identifying connections between papers. They can help locate additional literature not identified through the main systematic searches for the review.
Click on the plus (+) icons below to learn about some AI literature mapping tools.
ResearchRabbit
ResearchRabbit helps researchers explore, visualise, and track academic literature networks by suggesting related papers and authors in an interactive, AI-assisted interface.
ResearchRabbit is free.
ResearchRabbit sources data from OpenAlex and Semantic Scholar.
LitMaps
LitMaps allows users to create dynamic visual maps of academic literature, using AI to discover and track new relevant papers over time.
LitMaps is a freemium service, it's free to sign up. However, access to additional features comes at a cost to users.
LitMaps sources data from Crossref, OpenAlex and Semantic Scholar.
Connected Papers
Connected Papers builds a visual graph of related research papers based on co-citation and bibliographic coupling to help users quickly understand the landscape of a specific topic.
Connected Papers is a freemium service, it's free to sign up. However, access to additional features comes at a cost to users.
Connected Papers sources data from Semantic Scholar.
AI in screening software
Machine learning is a form of AI which is used in screening software. Typically, the AI learns from reviewers' decisions and ranks studies that are most likely to meet review selection criteria. Some screening software also include an AI feature that suggests when to stop screening.
Click on the plus (+) icons below to learn about some screening software that has AI functionality.
Covidence
Covidence uses AI to streamline systematic reviews by assisting with citation screening and data extraction, enhancing speed and accuracy.
Deakin has a subscription to Covidence for staff and students.
LitQuest
LitQuest is tool developed by Deakin that uses machine learning to expedite screening and includes data extraction and Living Review functionality.
LitQuest is free to Deakin staff and students.
Rayyan
Rayyan is a tool that uses AI in deduplication and screening.
Rayyan is a freemium service, it's free to sign up. However, access to additional features comes at a cost to users.
ASReview
ASReview is a open source software which uses AI for screening.
ASReview is free.