Default view
Artificial intelligence (AI) Searching with AI tools

Searching with AI tools

Web of Science – Smart Search (at the campus or with VPN)

Thanks to our university library’s subscription, students and faculty have free access to the Web of Science database, one of the world’s most respected scholarly citation and discovery platforms. The Web of Science Smart Search feature uses artificial intelligence to make research faster, more relevant, and easier—even if users are not yet familiar with all the keywords of a given field.

What can Smart Search do?

  • Natural language searching: you can ask questions in full sentences (e.g. “latest research on the economic impacts of climate change”).

  • More relevant results: the system interprets research intent, not just keywords (this still needs to be critically checked).

  • Identification of related topics: helps uncover new directions and connections.

  • Time saving: fewer refinements and a faster results list.

Searching smartly with artificial intelligence – prompting and more (

Chat-based artificial intelligence tools now offer multiple ways to search and use information. From the perspective of scholarly research, the most valuable tools are those that can process the most up-to-date information available online (although it is best when they search for us within a professional academic database).

Basically, there are two main approaches:

  1. Simple (prompt-based) searching, where we ask questions and receive a quick overview of a topic.

  2. Using AI agents, which perform more complex tasks: searching, filtering, summarising, and making recommendations in multiple steps.

These solutions effectively support orientation in the scholarly literature, especially when the goal is quick familiarisation, discovering new directions, and identifying relevant sources.

Artificial intelligence itself can also be used to create our own AI assistant.

Write a prompt to create a research assistant who searches for peer-reviewed scholarly articles in a university research group, is precise, and searches only for literature that is peer-reviewed and meets university quality guidelines, so that I can use it to build an agent.

The first response to the above prompt – which we can further refine ourselves before creating our AI assistant (e.g., by narrowing it to a specific discipline).

RESEARCH ASSISTANT (SCHOLARLY LITERATURE SEARCH)

You are a research assistant working for a university research group.
Your task is exclusively to search for and pre-screen peer-reviewed, high-quality scholarly literature.

Principles and quality criteria

You may only use peer-reviewed sources.
Accepted types include:

  • scholarly journal articles

  • peer-reviewed conference proceedings

  • systematic reviews and meta-analyses

You may not use:

  • blogs or popular science articles

  • predatory journals

  • unverified preprints (e.g. arXiv, SSRN), unless explicitly permitted

  • Wikipedia-style summaries

Sources must comply with university research quality guidelines, with particular attention to:

  • scientific reliability

  • transparent methodology

  • proper citation practices

Search requirements

  • Prefer recent publications (ideally from the last 5–10 years).

  • Use multiple search strategies (keywords, synonyms, conceptual networks).

  • Where available, prioritize by citation impact and journal quality (e.g. Q1/Q2 ranking).

Expected output

For each identified publication, provide:

  • Author(s)

  • Year of publication

  • Title

  • Journal / conference name

  • DOI or stable link

  • A short (2–3 sentence) summary of scientific relevance

  • A justification of why it meets peer-reviewed and university quality requirements

Operating principles

  • Be precise, critical, and conservative in source selection.

  • If the quality of a source is uncertain, exclude it.

  • If no suitable peer-reviewed literature is found, clearly indicate this.

Your goal is to ensure that the search results can be directly used for a university research project, publication, or grant application.

Short example prompts for searching

a. Search for international scholarly articles on non-surgical treatment options for canine hip dysplasia in golden retrievers.
b. Search for scholarly articles that define expectations regarding veterinarians’ communication skills, preferably from the 2020s onward; ideally, search separately for companion animals and farm animals.

What is important when evaluating search results?

When reviewing results, do not rely solely on the description provided by the AI or the search system. These can be useful starting points, but they do not replace critical reading of the source.

For a well-founded decision:

  • Read the abstract to gain a clear understanding of the research objectives, methodology, and results.

  • Then review the conclusions/summary, where the authors interpret their findings.

  • If you have a specific question, search the full text for the relevant keyword or concept.

  • Read the article.

  • Examine the relevant sections in detail to verify whether the article truly addresses the issues that interest you.

Only in this way can we be sure that the publication actually contains the information needed for our research—and that the AI’s claims about its content are accurate.

The decision and responsibility always remain in human hands. We do not passively “accept” the results we receive; instead, we actively shape the process by refining our questions, rephrasing searches, and critically evaluating outcomes.
Artificial intelligence does not make decisions for us—it is merely an assistant that helps us work faster, clearer, and more efficiently.