GenAI and AI-powered tools can be used to support literature searching by providing a range of exploratory search capabilities. These include providing a list of starting references, providing overviews of topics and brainstorming search terms for database searches. Some tools will also provide question answering, summarisation of text, and mapping connections between authors and between papers. Library databases may also include AI search features which enable natural language searching and provide summaries.
These tools are designed to support and enhance searching and research, not replace it. They should be used to complement expert search methods, comprehensive engagement with full texts and rigorous critical reading practices. For comprehensive literature searching, such as for a research paper or thesis, you should still search the major academic databases in your field.
The Library subscribes to a number of databases that include in-built AI functionality. These allow features such as Natural Language Processing (NLP) to allow users to type full questions or conversational sentences rather than complex Boolean strings. The AI parses the intent, extracts key concepts, and maps them to subject headings or keywords. Other features include suggesting additional search terms and providing AI-generated summaries of articles.
AI-powered tools in subscribed databases operate within closed vendor systems, meaning they do not accept user-uploaded content and do not access broader web resources. This can help ensure:
Some current limitations around these tools include:
Below is a list of current Curtin database subscriptions that include AI features:
ProQuest Research Assistant. See FAQs for more information.
Access ProQuest
EBSCOhost AI Companion tools. Includes AI-Assisted Search, AI Insights, Ask This Document (AI Ask) and AI Conversational Search.
Access EBSCO databases
JSTOR AI Research Tool. See FAQs) for more information.
Access JSTOR
CCH iKnowConnect AI Assistant. See FAQs for more information.
Access CCH iKnowConnect
Sage Research Methods AI Assistant. See FAQs for more information.
Access Sage Research Methods
JoVE AI Search. See FAQs for more information.
Access JoVE databases
General-purpose Large Language Model (LLM) tools, such as Claude and ChatGPT, and specialised GenAI tools can be useful for literature searching. These tools bring both benefits and risks. Curtin Library does not endorse use of specific tools – the tools listed below are in common use in the research community and the advice provided is general in nature. Always apply your own professional and academic judgement to assess tools based on your needs, responsibilities, and values. Apart from enterprise-licenced tools, Curtin does not provide access to paid versions of these tools, and some may require individual subscriptions.
What: ChatGPT and Claude are examples of GenAI chatbots powered by Large Language Models (LLMs). These are trained on a very large amount of data to predict and produce coherent text, answer questions, translate languages, write code, and perform various other language-based tasks.
Cost: Most GenAI chatbots are freemium tools with limited free functionality and additional functionality via subscription.
Data sources: LLMs are trained on a variety of content that comes from public web content, licensed or partner data, books, research, code, technical documentation, synthetic data and user data permitted for training. There is no single universal LLM data repository used by every model. Training data, retrieved documents and live web results also use different sources of model knowledge.
Uses: Can be most useful before starting a literature search to aid brainstorming and by clarifying concepts, topics and methodologies, and by generating and refining search terms.
Limitations: Outputs are derived from a restricted corpus that misses major subscription-only, paywalled or very niche scholarly articles because it lacks full-text access to many academic databases. Answers may not include authoritative sources or recent evidence, depending on when the model was trained and how often it is retrained or updated. ‘Hallucinations’ are still possible, including references that look plausible but do not exist, or that exist but do not support the claim being made.
What: Semantic Scholar is a free, AI-powered search and discovery tool. It provides exploratory semantic searching, citation searching and quick access to full texts.
Cost: Free to use, need to create an account to access all features.
Data sources: Uses S2ORC: The Semantic Scholar Open Research Corpus through the Semantic Scholar Public API. Also draws from open‑access sources, permitted web‑crawled PDFs, and partnerships with scientific journals, indexes and content providers.
Uses: Can be used for preliminary scoping searches and identifying potentially relevant literature. Offers AI features such as TLDR summaries, extracts key insights and identifies connections between studies.
Limitations: Does not support Boolean operators or wildcards. Lacks full transparency regarding publisher partnerships, licensing agreements and model training data. Over-reliance on summarised claims may discourage deeper reading or critical engagement with full texts.
What: Elicit is a semantic search tool that enables literature discovery, literature summarisation, and has research report automation and paper chat functionality.
Cost: Freemium tool with limited free functionality and additional functionality via subscription.
Data sources: Includes Semantic Scholar, OpenAlex and other publishers, repositories and preprint servers and government sources.
Uses: Can be useful for early-stage exploratory searches. Uses conversational-style search to provide summaries of papers, often highlighting study focus and methods.
Limitations: For many indexed papers, Elicit only has access to the abstract rather than the full text, potentially leading to a superficial understanding or incomplete data extraction. Advanced features, higher data-processing capacities and full integrations require paid subscriptions. Over-reliance on summarised claims may discourage deeper reading or critical engagement with full texts.
What: Consensus is an AI-powered search tool designed to support exploratory, question-led research.
Cost: Freemium tool with limited free functionality and additional functionality via subscription.
Data sources: Includes publicly available abstracts via Semantic Scholar, OpenAlex, and additional academic content aggregated through its own data collection and publisher partnerships.
Uses: Can be useful at the start of research to discover relevant studies and evidence related to specific claims. Uses semantic search and Natural Language Processing (NLP) to extract, summarise and assess consensus across academic literature, drawing on abstracts and, where available, selected full-text content.
Limitations: Full text analysis is dependent on publisher partnerships and is not consistently applied across all indexed literature, limiting access to context and methodological details. Lack of explanation of how “consensus” is measured or calculated, which can lead to misinterpretation. Over-reliance on summarised claims may discourage deeper reading or critical engagement with full texts.
What: Undermind is an AI-powered research discovery and literature review support tool.
Cost: Freemium tool with limited free functionality and additional functionality via subscription.
Data sources: Main source is Semantic Scholar.
Uses: Can be useful to provide topic-driven searches, classifying relevance of papers and synthesising summaries. Users can start with a research question, use prompting and obtain a report outlining key literature.
Limitations: Source set is not comprehensive and should be used in combination with other sources. Over-reliance on summarised claims or chat function may discourage deeper reading or critical engagement with full texts.
What: ResearchRabbit uses AI-powered algorithms to generate visual networks of scholarly literature and recommend scholarly articles and authors based on citation, co-citation, and co-authorship networks.
Cost: Free to use but must be logged in. Subscription available for ResearchRabbit+.
Data sources: Primary data sources are Crossref, Semantic Scholar and OpenAlex.
Uses: Can be useful as a starting point for searching, environmental scanning and staying updated on emerging research. Also useful for citation mapping and visual discovery.
Limitations: Mandatory registration required before the tool can be used. Search outcomes can vary based on input method (title vs. DOI).
What: LitMaps is an AI-powered citation discovery tool that supports literature discovery and visualisation through citation networks. It builds a map of inter-related papers based on shared references, citations and co-citations of seed articles and locates related articles using AI ‘similar text’ algorithms.
Cost: Freemium tool with limited free functionality and additional functionality via subscription.
Data sources: Includes open access metadata from Crossref, Semantic Scholar and OpenAlex.
Uses: Can be useful to complement literature searching, from scanning the broader literature for a particular topic to identifying seed articles and key authors. Also useful for citation mapping and visual discovery.
Limitations: Features that allow users to refine or precisely structure searches are only available to premium users.
Note: Some of the content included here was derived from Deakin University Library’s detailed evaluations of AI tools. For more information see AI Evaluations – Library.