Artificial intelligence (AI) is increasingly changing literature search. In addition to generative systems such as ChatGPT, which can assist with writing or brainstorming, there are specialised AI-supported tools that specifically search scientific publications and reveal contextual connections.
You can find more information about the individual tools in the overview AI-Tools for Literature Search.
Prepare Literature Search – Develop Topics and Search Terms
Generative AI can be a useful tool in the early stages of your research before you switch to scientific databases. Use generative AI to
- narrow down topics or develop new, more specific questions,
- identify synonyms, related terms, and alternative spellings for later searches in databases (e.g. Web of Science, Scopus, PubMed).
Tools: ChatGPT, Gemini, Copilot, Perplexity etc.
Do not use generative AI to search for scientific sources directly – the information may be incomplete or incorrect. Furthermore, do not enter any personal or confidential data into such systems.
Searching for Literature – Identifying Relevant Sources
Specialised AI tools enable you to search for real, scientific sources. While traditional databases are primarily based on keyword matches, many AI-supported tools work semantically: they capture the meaning of a question and thus deliver broader or new results. They often use the principle of retrieval-augmented generation (RAG) – a combination of database search and language model. This allows them to recognise contextual connections that go beyond mere word matches.
Tools: Semantic Scholar, Scite, Consensus, Elicit etc.
Mapping Literature – Visualising Connections
If you have already found relevant works, AI-supported mapping tools can help to visualise research landscapes and identify connections between publications. They link publications, show thematic clusters and visualise the development of topics over time and across authors.
Tools: Open Knowledge Maps, Research Rabbit, Connected Papers, Inciteful etc.
Limitations
AI-supported tools can open up new perspectives, but they cannot replace traditional, systematic literature searches. The conscious and critical handling of results, including checking, documenting and citing correctly, is an integral part of good scientific practice.
- Accuracy & quality: Results should always be checked and verified for citability. If an AI tool does not display full texts, access to the relevant sources should be ensured via library access or interlibrary loan.
- Database: Most tools only search open access material and abstracts, not full texts. Licensed content is not searched. Books and printed works are rarely covered.
- Transparency & reproducibility: It is often unclear why certain sources are displayed or how the results are sorted. Similar search queries can lead to different results; systematic, repeatable searches are hardly possible.
- Bias: Search results reflect the focus of the training data – certain regions, languages or disciplines may be underrepresented.
Not sure how to use AI effectively in your research process?
- In our blog post, you can learn how to use AI step by step in the various phases of literature searches – with example tools and prompts.
- Our overview AI-Tools for Literature Search provides an up-to-date compilation of AI-supported research tools with functions, possible applications and links.
- In our workshops, you will learn how to use AI efficiently and critically in your research. Dates and registration details can be found on the library's events page.