Universität Bonn

University and State Library

Using AI for Literature Searches

Thanks not least to ChatGPT, machine learning and generative artificial intelligence (AI) are becoming increasingly important, including when studying and learning. This page provides information and advice designed to help you use AI during your studies, specifically in your literature searches.

What is AI and how does it work?

The term “artificial intelligence,” or “AI,” refers to computer-based systems that perform tasks requiring human skills such as understanding, learning and decision-making.

Rather than “thinking” like a person does, it uses mathematical models to process information and is often based on machine learning. Vast quantities of data are used to train it to spot patterns and make predictions.

For example, the kind of language-based AI that is also used in literature searches will calculate the probability of certain content appearing in a certain order.

However, it is important to remember that AI does not possess independent knowledge or have a genuine understanding of truth or of academic or scientific quality. Its results are based on probabilities and may contain errors.

For research work, AI is chiefly useful as a tool for getting a handle on vast quantities of information, giving topics a structure and devising new search strategies. Users should still be the ones to appraise sources and retain responsibility for their research.

COLOURBOX66687448.webp
© colourbox.de, ID: #66687448

AI can...

  • spot patterns in texts;
  • suggest relevant literature;
  • summarize topics;
  • compile lists of search terms; and
  • provide support with questions.

AI cannot

  • appraise scientific or academic quality itself;
  • form its own opinions or knowledge;
  • be guaranteed to supply correct answers; or
  • take responsibility for source work.

Good to know

AI is a tool designed to supplement research, not replace it. Academic or scientific appraisals remain the responsibility of the user at all times.

Using AI for literature searches

AI systems used for literature searches work on the basis of machine learning and statistical models. Unlike traditional search engines, which merely search for key words, AI systems analyze vast amounts of text, spot connections and produce suggestions that might be relevant in terms of their content.

Especially in the preparatory phase, AI can be used to support literature searches when

  • identifying a topic and narrowing your research focus;
  • identifying key words and search terms; and
  • producing overviews or summarizing large amounts of text.

How AI works in literature searches

  1. Analyzing literature data: AI will analyze bibliographical data such as titles, abstracts, key words, citations and—depending on the specific tool—full texts from academic and scientific publications.
  2. Spotting thematic links: AI will use this data as a basis for identifying research topics, key terms, related concepts, content overlaps between publications and links between citations.
  3. Identifying relevant literature: AI will suggest suitable sources and arrange them by topic. 
  4. Adapting interactively: These results can be refined by making search queries more precise, asking follow-up questions or applying corrections.
COLOURBOX63442004.webp
© colourbox.de, ID: #63442004

When used in searches, AI can…

  • help to find relevant literature;
  • highlight topics and connections;
  • propose key words and search strategies; and
  • provide overviews and summaries of large amounts of text.

When used in searches, AI cannot…

  • guarantee that all results will be correct or complete;
  • appraise scientific or academic quality;
  • replace the power of human judgment or critical thinking.

Good to know:

AI calculates probabilities, not absolute truths. Although it can recommend relevant literature, it also has the potential to make mistakes or “hallucinate” content that does not exist.

Library search portals, catalogues and specialist databases are still vital components of a reliable search for academic and scientific literature.

Various types of tool

AI tools can be grouped based on what they can do. 

Rather than search terms, many AI tools respond to prompts, i.e. commands worded in everyday language.

Help with

  • brainstorming; 
  • formulating search strings for database queries;
  • analyzing texts;
  • wording outline points;
  • writing first drafts; and
  • proofreading ideas after committing them to paper.

Help with 

  • analyzing documents;
  • explaining specific parts of a text;
  • identifying fundamental ideas and methods; and
  • writing summaries.

Help with 

  • searching large databases for relevant literature;
  • finding recent and relevant academic articles; and 
  • providing additional features such as citation and analysis tools (in many cases).

Help with 

  • identifying thematic links between academic papers;
  • illustrating these links in visual form;
  • discovering new relevant publications based on a starting paper; and
  • understanding connections between literature sources.

Six questions to ensure the considered use of AI

Be aware that using AI means availing yourself of a “third-party service.” You should therefore ask yourself:   

  • Do I want to use an AI tool—and, if so, what for?
  • What are the advantages and disadvantages of working with an AI tool as far as I am concerned?
  • Am I even allowed to use AI tools in my department
  • Is the AI tool suitable for research work?
  • Do I have the knowledge required to use the tool?
  • How can I check that the results of my work with the AI tool are correct and reliable?
COLOURBOX61789843.webp
© colourbox.de, ID: #61789843

N.B.:

AI must be used mindfully and transparently in research work.

This use should always comply with the regulations of the relevant university, department or examination. Personal or copyrighted content must not be fed indiscriminately into AI systems.

You should therefore consider

  • the principles of its use in a research context; 
  • transparency and disclosure requirements; and
  • the citability of AI-generated content.
  • Handling copyright, licenses, privacy and data protection
  • Documenting AI use

Legal aspects

Whenever you use AI, please bear in mind that most AI providers do not feel the need to comply with German or European legal standards governing the protection of personal data or copyright regulations, so you should always obey three rules:

  • Do not enter any personal data of third parties into AI applications, and think carefully before disclosing any of your own.
  • Do not upload any copyrighted texts into AI applications (not even extracts).
  • If you are intending to use AI, make sure that this will be permitted under the examination regulations or equivalent instructions applicable to your subject.

AI and the University of Bonn

The Guiding Principles of the University of Bonn on Handling Generative Artificial Intelligence in Teaching and Learning (in German) were published in October 2025. The basic principles that they set out for handling AI in University teaching and examinations—while taking account of privacy, data protection, copyright and disclosure requirements—are designed to provide guidance to all associates of the University.  

Because different disciplines face different requirements and challenges when it comes to using AI, the faculties are called on to consider these guiding principles in light of the specific requirements of their subjects, adjust them if need be and publish their final versions. 

All sets of principles will be reviewed and updated on an ongoing basis.

The University of Bonn is providing a number of key handouts, training courses and an internal information platform in order to aid efforts to put these sets of guiding principles into practice: KIWIS (“Künstliche Intelligenz: Wirkungsvolle Integration von Sprachmodellen,” or “Artificial Intelligence: the Effective Integration of Language Models”).


Conclusion

Rather than being a guaranteed pathway to success, the use of AI requires active intellectual assistance from the user.

They must:

  • Understand AI: You must be familiar with how AI tools work and what risks they pose. Achieving the desired results takes practice.
  • Apply the requisite specialist knowledge: You will need a basic understanding of your subject in order to gauge the quality of the results presented to you. This is the only way to
    • assess the credibility of your results;
    • detect any bias; and
    • subject your results to critical scrutiny and use them responsibly.

AI is a tool, not an authority!


Contact

Information desk Main Library

We're here to help with any questions you may have about the library or literature research: Monday through Friday, 9 a.m. to 5 p.m.

Contakt

+49 228 / 73-7525

information@ulb.uni-bonn.de

Wird geladen