Skip to Main Content
Turku University Library

Nursing Science

AI in information retrieval

Generative artificial intelligence (GenAI or GAI) is becoming more common in both everyday life and in academic setting. There are many different types of AI tools and they can be used for different purposes.

Tools available to university staff and students for the academic year 2024-2025

The AI tools available through Turku University Library are available to you with your UTU ID. There is no need to create a separate user account for the services.

  • Microsoft Co-pilot is a chat application that generates text and images. The app is connected to the internet. It provides links to the websites it has used to generate the responses. When the application is used with a logged-in UTU account, chat data is not stored, Microsoft cannot view the person's or organisation's data, and the data is not used to train AI.
  • Keenious searches for similar articles based on a given example article or text. University students have access to a protected version, which does not store usage data and is not used to train the program. PDF files of articles can be entered into the program.
  • Scite.ai has a many different functions for finding articles and managing references. You can search for articles by title, author name, subject or DOI. The Assistant generates a summary of the topic based on a query and searches for related sources. Reference Check analyses your document and finds references based on the content of the document.
  • ScopusAI A tool linked to the Scopus database where you can enter a search term in natural language (similar to Googling). The tool generates a search based on the given topic and generates a summary of the topic, searches for archives on the topic, generates a concept map and identifies key researchers in the field.
  • Web of Science Research Assistant presents the answers in chat format. You can enter a question or topic in the same way as for ScopusAI and receive a summary of the topic and article recommendations. The Research Assistant also generates a concept map of the topic and identifies key experts in the field.

AI applications for article search

For article retrieval, it is better to use dedicated applications based on databases that index real articles. Depending on the application, articles can be searched by keywords, article title, identifier (e.g. DOI), self-written text, etc. However, when using these applications, take into account copyright restrictions on the use of articles. Some publishers have stated that the materials they publish may not be used to train AI. This means, for example, that a PDF file of an article may not be fed into an AI application.

Apps are not a substitute for systematic information retrieval, but are a good complement to it.

ChatGPT and other major language models

Large Language Models (LLMs) based chat applications such as ChatGPT generate text based on their training material. With them, conversations are at best natural and responses can be credible.

However, the language model does not know whether the answer it gives is right or wrong, but generates probabilities by calculating the most appropriate answer. Furthermore, not all applications are connected to the internet, so they do not have access to the latest information.

ChatGPT and other applications based on language models can be used, for example, to come up with search terms. However, they should not be used for searching for sources, as there are many cases where a language model has generated a source reference that looks plausible but does not actually exist.

 

More information on Librarian's guide to Artificial Intelligence