Search tips

Searching for literature and scholarly sources can be time-consuming. This page provides guidance on effective search strategies, the characteristics of scholarly articles, and how to evaluate the credibility of sources.

Prepare your search

Before you start searching for information, you need to consider the purpose of your search. Do you want to get an overview of a topic or explore a specific question in depth? The purpose determines how you should search.

If you are conducting an exploratory search and do not need to report how you searched, you can work more freely. If the search needs to be reproducible, it must be systematic. Start by considering the following:

• Why are you searching for information? Do you need an overview or an in-depth study?
• What type of material do you need, for example books, scholarly articles, legislation or statistics?
• How much material do you need?
• How current does the material need to be?
• Does the material need to be scholarly?

Systematic search

A systematic search is carried out in a planned and structured way according to a defined method. The aim is for the search to be transparent, meaning it should be clear how it was conducted and which sources were used.

The search must also be reproducible. This means that another person should be able to repeat the search and obtain similar results. It should also be comprehensive, meaning that it includes all relevant articles that can answer the research question.

Exploratory search

An exploratory search does not need to be documented or reported. You can use it at the beginning of your work to:

• Identify a research problem.
• Learn more about the subject area.
• Determine whether there is sufficient research in the field.
• Narrow or refine your search question so it is neither too broad nor too narrow.

Documenting the search

When conducting a systematic search, you need to document all parts of the process, since the search must be transparent and reproducible. Many databases have a feature called Search History, where you can find and save your previous searches and filters.

In addition to the search results, you should document:

• When you conducted the search.
• Which databases and search tools you used.
• Which search terms and search techniques you used.
• Which filters you applied, for example peer reviewed, language or publication year.
• The number of results in each database.
• How you selected the articles.

How do I find search terms?

Finding good search terms is often challenging, but it is an important part of the search process. There are many tools that can help, such as encyclopedias and dictionaries. You can also use key articles and subject term lists.

Start with your research question and identify the words that describe the most important concepts. These words form the basis of your search. Then consider whether there are synonyms, broader terms or more specific terms you can use. Scholarly material may use different words for the same concept. By varying your search terms, you increase your chances of finding relevant results.

Scholarly texts are often written in English. You therefore need to translate your search terms and use the English terms in your search.

To find good search terms, you need to learn more about your topic. Use course literature, lectures, websites and encyclopedias to understand which concepts are central. You can also use AI search tools such as Microsoft Copilot or ChatGPT to get suggestions for search terms and synonyms. Use them as support, but always check that the suggestions fit your research question.

• Encyclopedias and dictionaries
  Encyclopedias can provide basic knowledge about your topic and help you identify relevant search terms.
  

Key articles

If you have found a scholarly article that is relevant to your work, you can use it to find more articles on the same topic. This is called using a key article or a gold standard article.

You can find new search terms in the article’s title, abstract and in the author’s keywords. In some databases, controlled subject terms are also linked to the article. Use these terms in your continued search to find similar material.

Controlled subject terms

Some databases include subject term lists with controlled vocabulary. This means that the database has decided which term should be used for a particular concept. These terms describe the main content of an article.

The subject terms are organised in a hierarchy. Broader terms appear higher up, and narrower terms further down. The database uses these terms to describe and organise scholarly articles and other material. They may be added manually by a person or automatically by the system.

When you search using controlled subject terms, you retrieve material that has been indexed with those specific terms.

Subject term lists may have different names and include different terms depending on the database. They may be called a thesaurus, subject headings or an index.

Where should I search?

Where you search depends on the type of material you need and the purpose of your search. For example, you can use the Library Search Tool or subject-specific databases.

Start by defining what type of material you need. Do you need printed books, e-books or scholarly articles? Should your search be systematic or exploratory?

Search systems differ in content, subject coverage and transparency. Transparency refers to how clearly it is explained how the search works, which sources are included and how the results are selected.

The Library Search Tool

The Library Search Tool is a good starting point. It allows you to find books, e-books and articles in one place. When you use the search tool, you search multiple databases and indexes at the same time. This often results in many hits. To get more relevant results, you need to refine your search using filters.

Printed books and e-books

Use the Library Search Tool to find books. Start by searching and then limit your results by selecting the source type (books or e-books). The search is automatically limited to books and e-books available at the University Library at Luleå University of Technology.

If you want to borrow a book from another library, you can request it through interlibrary loan. Please note that e-books cannot be borrowed via interlibrary loan.

• Books and e-books
  Information on how to search for and borrow books and e-books, including access and loan conditions.
• Borrowing
  Information on how to create a library account and about loans, renewals and interlibrary loans.

Articles

You can search for articles in the Library Search Tool by entering keywords, a title, an author or an ISSN. When you use the search tool, you search multiple databases and indexes at the same time. This often results in many hits.

To reduce the number of results, use filters such as Peer reviewed, language or the source type Academic journals. If you still get too many results, it may be better to search in a subject-specific database.

• The Library Search Tool
  Search for books, e-books and articles here.
  

Subject-specific databases

At the University Library, you can access many different databases. Some cover several subject areas, while others focus on a specific field.

A database is a collection of searchable material, such as scholarly articles. Some databases provide full-text access. Others contain only references to sources. They may also include data such as statistics, standards and patents.

The library’s database list includes both open-access databases and subscription-based databases. Databases that require a subscription are marked with (LTU). To use them, you need to log in with your LTU account.

The advantage of searching in a subject-specific database rather than in the Library Search Tool is that you can apply more and more specific filters. For example, you can limit your search by age group, geographical area or evidence-based material. In many cases, you can also use controlled subject terms in your search.

Keep in mind that databases function in different ways. You should therefore read the database’s help section to learn more about how to search effectively.

• Databases
  Here you can find the University Library’s databases. They are listed alphabetically or by subject.
  

Google

If you search using Google, you may miss a significant amount of scholarly material. It is therefore often better to use the University Library’s search systems and databases.

You can also use Google Scholar to find scholarly material. However, not all material in Google Scholar is scholarly, and it is unclear how the selection is made. It is therefore important to evaluate the sources critically. The content in Google Scholar is not permanent. This means that articles you found previously may no longer be available the next time you search.

• Google Scholar
  When you are logged in at Luleå University of Technology, it shows which material is available in full text.
  

AI search tools

There are several types of AI tools that can be used for information searching. They work in different ways and are not suitable in all situations. When using such tools, it is important to understand how they function so that you can use them safely and responsibly.

You cannot use AI search tools when the search must be fully transparent and reproducible. This is because you cannot examine exactly how the tool has processed the query. You also cannot recreate the same search, even if you use the same wording.

Generative AI

When you search for information using AI, generative AI is often used. It is based on a technology called machine learning. This means that the computer learns from data instead of being programmed for each task.

The purpose is to allow the computer to handle complex tasks automatically. This is done by using algorithms and statistical models. These models identify patterns in large amounts of data. When a model is trained with more data, the results usually improve over time.

Deep learning and neural networks

Deep learning is a more advanced form of machine learning. It uses deep neural networks to process information. A neural network is a mathematical model inspired by how the human brain processes information. It is built from several layers. Each layer contains nodes (or neurons) that handle information step by step. Neural networks can be used for many different tasks. When they are used to create new content, such as text or images, it is called generative AI.

Large language models – LLM

There are different types of generative AI. One example is large language models, often called LLMs. Large language models are a type of neural network based on a technology known as transformer architecture. These models are used to understand and generate text that is similar to human language. They are commonly used in chatbots and AI-based search tools.

Examples of large language models are:

• GPT
  GPT was developed by OpenAI. It is used in services such as ChatGPT and Microsoft Copilot. The model can answer complex questions, reason about problems, write code and generate text.
• Claude
  Claude was developed by Anthropic. It is used in services such as Claude.ai and Notion AI. Claude can write and summarise content. It is especially good at producing longer texts, summaries and well-structured responses.
  
• Gemini
  Gemini was developed by Google DeepMind. It is used in Google AI services and in Google Workspace tools such as Gmail and Google Docs. The model can handle different types of information at the same time, including text, images and video.

Personal privacy and data security

When using AI tools, you are responsible for ensuring that you use them safely, both for yourself and for others. Assume that any information you share may be stored, shared or used to train and develop AI models.

Therefore, you should never share sensitive information. This includes personal identity numbers, addresses, passwords, payment card details or confidential information. Sensitive information may also include audio recordings or transcripts from meetings.

In addition to the information you share, AI tools may collect other data, such as your IP address, email address and location data.

• The Swedish Authority for Privacy Protection – IMY
  IMY describes the risks to personal privacy when AI is used. Available in Swedish.
• Use generative AI in an ethical manner
  Digg, the Swedish Agency for Digital Government, provides information on how you can use generative AI in a responsible and ethical manner. Available in Swedish.
• Ethical aspects
  Who is behind AI tools, and are there other ethical aspects to consider? Karolinska Institutet University Library clarifies the concepts.
• What does AI do with your data?
  Internetkunskap from Internetstiftelsen explains what happens to your information when you use AI tools. Available in Swedish.

Chatbots

ChatGPT and Microsoft Copilot are examples of chatbots based on the GPT language model. They generate responses based on the text you enter in your question, also known as a prompt. The responses are based on statistical calculations of what is most likely to follow your input. For this reason, the same prompt may produce different answers at different times.

It is not publicly known which material the models have been trained on. It is also not possible to determine which sources an individual response is based on.

Some chatbots use a technique called Retrieval Augmented Generation, RAG. This combines the model’s ability to generate text with searching in external sources, such as websites or documents. Whether this function is used depends on the settings and on what external data the tool can access.

If you need to find scholarly sources, you should not rely on chatbots. The sources behind the answers are often not visible, so you cannot tell where the information comes from. Even if you ask for sources, chatbots may generate references that do not exist. Therefore, it is important to always verify information in other reliable sources.

However, chatbots can be useful at the beginning of your work, when your search is more exploratory and experimental. For example, you can use them as a discussion partner to:

• Get started with an assignment.
• Test ideas or gain inspiration.
• Generate possible research questions.
• Translate words from Swedish into English.
• Find search terms and synonyms.

You can also ask a chatbot to create search strings for different databases. Keep in mind that the results often contain errors. For example, important search techniques such as phrase searching and truncation may be missing. The quality of the search string also depends on how you formulate your question.

• A practical introduction to generative AI – ChatGPT, Gemini, and Copilot
  Digiteket offers an introductory course on generative AI. Available in Swedish.
• Examples of AI tools that serve different purposes
  Tips on useful chatbots from the Karolinska Institutet University Library.
  
• Microsoft Copilot – AI service for students
  Learn more about what is included when you use Microsoft Copilot for students.
  

AI search tools for scientific material

There are AI search tools that retrieve actual sources based on your query and provide answers grounded in scholarly publications. They also display the sources used. The technology behind these tools is called Retrieval Augmented Generation, RAG. It combines the language model’s ability to generate text with searches in databases or search engines.

The purpose of this technology is to provide more fact-based answers and reduce the risk of so-called hallucinations, that is, fabricated or incorrect information. It also makes it possible to use information that the model has not been trained on. This allows the tool to answer questions about specific documents.

Elicit and Perplexity are examples of AI search tools that use RAG. They may be connected to general search engines such as Google or to research-oriented services such as Semantic Scholar.

One limitation of these tools is that they often do not have access to full-text articles. Instead, their answers are based on abstracts or metadata. This may mean that important details are missing. The quality of journals varies, and it is not always clear why certain sources are selected over others. This may result in biased or incomplete answers.

Agent-based AI tools

There are also other types of AI search tools, such as agent-based tools. These work step by step to complete a task. If you ask a question, the tool may first rephrase it. It may then generate its own prompts to retrieve relevant references. The exact process varies between tools.

Snowball search AI tools

Snowball search AI tools are another type of tool. They are not generative. Instead, they identify articles based on studies you already have, for example by analysing reference lists, citations or similar content. This method should not be confused with snowball searching in databases such as Web of Science and Scopus, where the search is based on verified citation chains.

• Examples of scientific AI search tools
  Karolinska Institutet University Library has compiled a list of AI tools that focus on scholarly material.
  

Subject guides

If you are unsure where to begin your search, you can use the library’s subject guides. They include databases, journals and other resources within different subject areas. You will also find links to additional information.

• Subject guides
  Here you can find subject guides for engineering, economics, history, law, the arts, medicine and health, education, psychology and sociology, as well as languages and literature.
  

How should I search?

Once you have identified the purpose of your search and where to search, the next step is to consider how to search. Scholarly databases, for example, do not function in the same way as generative AI tools. It is therefore important to understand how the systems are structured so that you can decide which one best suits your purpose.

Lexical and semantic search – what is the difference?

When you use a traditional database, without its possible AI features, you are usually performing a lexical search. This means that the database looks for exact words or phrases in fields such as the title, abstract or subject headings.

AI search tools, on the other hand, use semantic search. This means that the search is based on the meaning of your query. The tool looks for content that is similar to what you are asking about, even if the words are not exactly the same.

As mentioned earlier, AI search tools can be used when the search is more exploratory. They can also be used as a complement to a more systematic search in a traditional database. However, when the search must be fully transparent and reproducible, you should not use AI search tools, since it is not possible to review or recreate the search process.

• AI and information retrieval
  Learn more about the difference between lexical and semantic search on the KTH Library website.
• Difference between traditional databases and AI search tools
  Table comparing library databases with AI search tools. From Karolinska Institutet University Library.

Common search techniques in databases

When searching in a database, especially during a systematic search, you need to use search techniques. These determine how the database interprets your query. Not all search techniques work in every database. Therefore, use the help section in the database you are working with to check what applies.

Phrase searching

If the words in your search need to appear in a specific order, you can place them within quotation marks, for example “Second World War”. The database will then search for the words as a single phrase in the exact order you entered. If you do not use quotation marks, the database searches for each word separately. This usually gives more results, but many of them may be less relevant.

Truncation

A word can have different endings. To include all variations in your search, you can add an asterisk after the word stem. This is called truncation.

Example: school* retrieves words that begin with school, such as school, schools and schooling.

If you do not use truncation, you will only retrieve results with the exact word form. For example, searching for school will not retrieve results containing schools or schooling.

Boolean searching

To combine multiple search terms, you can use the Boolean operators AND, OR and NOT.

• AND is used when all search terms must appear in the same article. It reduces the number of results. For example, television AND movies retrieves documents where both words appear.
• OR is used when at least one of the search terms must appear. It increases the number of results. OR is useful when searching for synonyms or spelling variants. A search for television OR movies retrieves documents containing television, movies or both.
• NOT is used to exclude a term. It reduces the number of results. A search for television NOT movies retrieves documents that contain television but not movies. Use NOT carefully, as you may exclude relevant results where both terms appear.

When combining synonyms with OR, use parentheses to show how the search should be interpreted.

Example: (children OR kids) AND television. The results will contain children or kids, together with television.

Controlled subject terms

Some databases provide subject term lists with controlled vocabulary that you can use in your search. A controlled subject term is a standardised term chosen to describe a specific concept. This means that the same term is used consistently for the same content.

To search using controlled subject terms, you first need to find the correct term in the database’s subject term list and then select it. If you do not, you will be searching using free text instead.

Use the database’s help section for more information on how subject terms work. Also note that the terms may differ between databases.

Field searching

Field searching is an advanced search method used to narrow down results. In a database, each publication contains specific fields such as author, title, publication year and keywords. You can use these fields to control where the search is performed, for example in the author, title or abstract field.

Citation chaining

By reviewing reference lists in publications, you can identify new sources for your work. This is called manual searching or citation chaining. In citation databases such as Scopus and Web of Science, you can also see which publications have cited a particular source.

Searching with natural language in databases

Many databases include features that allow you to use AI in your search. In the library’s search tool, EBSCO Discovery Service (EDS), for example, there is a feature called EDS Natural Language. This allows you to formulate your query in everyday language, similar to asking a person a question. You do not need to use specific search terms or combine them with Boolean operators.

Examples:

• Which articles discuss AI in education?
• How does climate change affect the Arctic?

When a database uses natural language, it automatically creates a search string based on your question. You can often view how the search string has been constructed. You may then copy, save or use it in another database.

Automatically generated search strings often lack important search techniques, such as phrase searching with quotation marks or truncation with an asterisk. The search may also become broad and produce many irrelevant results.

Similar features are available in other databases, such as CINAHL, APA PsycInfo, ERIC and Web of Science. Many databases also include other built-in AI features. Some of these require a separate subscription.

If you are conducting a systematic search, you should not use search options such as Natural Language or Smart Search. You should also avoid features where the database automatically adds related or equivalent terms or subject headings.

Limiting your search in databases

In most databases, you can limit your search results by applying filters, such as publication date, language or source type. This helps reduce the number of results and identify the most relevant material. You can also use the filter Peer reviewed or Peer reviewed journals to display scholarly reviewed material. Always verify that the material has actually been peer reviewed.

Searching with AI search tools

When using AI search tools, you formulate instructions in everyday language, similar to asking a person a question. These instructions are called prompts. How you phrase your prompt affects the answer you receive. It is therefore important to think carefully about how you express yourself.

Writing effective prompts

A good starting point is to provide the tool with clear background information. Describe who you are, who will use the material and in what context it will be used. This often results in more useful responses.

After your initial prompt, continue step by step. You can refine or adjust your instructions after each response. This process is called iterating, meaning that you gradually improve your prompt to achieve a better result. Avoid leading questions and try to remain as neutral as possible. Otherwise, you may influence the answer and introduce bias.

• Prompt like a pro – how to improve your AI instructions
  Examples of how you can formulate better prompts when using AI. Created by Internetkunskap from Internetstiftelsen. Available in Swedish.
• Prompt tips for AI search tools
  Karolinska Institutet University Library provides examples of how to write clear and effective prompts in AI search tools.
• What is prompting?
  Digiteket explains how prompting works and what you need to consider when writing instructions for AI tools. Available in Swedish.

What sources should I use?

A source is the material from which you obtain information. It can include conversations, handwritten letters and blogs, as well as journals and scholarly articles. Which sources are appropriate depends on the purpose of your work and how you intend to use them.

Sources vary in their level of scientific basis and how they are used. At the beginning of a project, it is common to use several types of sources, such as textbooks, review articles, dissertations, brochures and popular science articles. These can provide an introduction to the topic and help you formulate a research question. In theses and other scholarly texts, scholarly articles need to form a central part of your material.

Sources are often divided into three categories:

• Primary
  A primary source is the original source in which information or research findings are published for the first time. It contains original data.
• Secondary
  A secondary source presents or interprets the content of a primary source. If you use a secondary source, it is important to locate and review the primary source as well.
• Tertiary
  A tertiary source summarises or compiles information from secondary sources. It is useful for gaining an overview of a subject.

What is scholarly publishing and peer review?

When researchers publish their findings, it is called scholarly publishing. This mainly takes place through articles in scholarly journals. A scholarly journal should contain primary research, meaning new research. Such publications are considered primary sources.

Research can be conducted in different ways and may vary in scope. Common types of studies include systematic reviews, randomised controlled trials, non-randomised studies, cohort studies and case studies. The reliability of studies depends on how they are designed, conducted and analysed.

To ensure quality, articles are reviewed in a process called peer review. This means that other researchers evaluate the journal before it is published. An article that has undergone peer review is considered scientifically reviewed. In some databases, you can filter results by peer review. In others, all material is already quality reviewed. In such cases, it is important to limit your results to scholarly articles if you are looking for primary sources. You must still assess the relevance and credibility of each source yourself.

• What characterises a systematic review?
  Here you can find more information about systematic reviews and how to evaluate the reliability of articles.

Different types of research publications

Research can be published in formats other than scholarly journal articles. Some types of publications are peer reviewed, while others are not. Here are some examples:

• Conference proceedings
  When researchers participate in conferences to share ideas and results, their contributions are often published in scholarly journals or in conference proceedings. In some cases, these are peer reviewed.
• Research reports
  Researchers may publish their findings in reports issued by departments, research institutes, organisations or government agencies. The purpose is to present ongoing or completed research. These reports are usually not peer reviewed and are often published freely online.
• Licentiate and doctoral theses
  After completing doctoral studies, researchers present their work in a thesis. A thesis is a scholarly publication in which the research is presented for the first time. It is academically reviewed and usually published openly online.
• Books and book chapters
  In many disciplines, research is published in book form. Books may be peer reviewed in the same way as scholarly articles. In other cases, they are reviewed by editors.

Grey literature

Grey literature refers to material that is not published through traditional publishers or scholarly journals. It is often produced by universities, government agencies, organisations or companies. Examples of grey literature include theses, conference papers, research reports, government reports, policy documents, clinical guidelines, working papers and information about ongoing research.

Grey literature is usually not peer reviewed. This means that the quality may vary and that the material may sometimes contain weaknesses or incorrect conclusions. At the same time, some reports may have been reviewed internally or by experts, even if they have not undergone a formal peer review process.

Grey literature may be found in subscription-based databases. For example, some conference papers are available in Web of Science, and theses are often published in university repositories. You can also find grey literature through search engines or on the websites of organisations. There are also specialised databases that collect this type of material.

Grey literature is mainly relevant for researchers, who may need to identify ongoing or unpublished studies. It is particularly important in emerging research areas where few studies have yet been formally published.

Student theses

Are student essays considered scholarly publications? A bachelor’s or master’s thesis may be structured like a scholarly article and follow the IMRaD model. IMRaD stands for Introduction, Methods, Results and Discussion.

Despite having a scholarly format, a thesis at undergraduate or master’s level is not considered a scholarly publication. It usually does not contain primary research, meaning new research, and this is not required at that level.

Source criticism

Source criticism means reviewing and evaluating the sources you use in order to assess their credibility. A critical approach is an essential part of academic writing and scholarly work.

According to the Swedish Higher Education Act (1992:1434), your work must be based on a scientific foundation. This means, among other things, that you must be able to find and use scholarly texts and research findings. The information should be reliable, reviewed and research-based. Theses and other scholarly works should primarily be based on scholarly articles.

• The Swedish Higher Education Act (1992:1434)
  The full text of the Higher Education Act. Available in Swedish.

You therefore need to examine and evaluate the sources you intend to use, as they may vary in credibility. How you assess a source depends on the type of source and how it will be used.

Start by identifying what type of source it is. Is it a review article, a dissertation or a scholarly article? Is it a primary, secondary or tertiary source?

When evaluating a source, consider the following:

• Authorship
  Who is the author or publisher? What qualifications and authority does the author have?
• Purpose
  What is the purpose of the material? Is it intended to inform, persuade, spark debate or something else?
• Audience
  Who is the material intended for?
• Currency
  When was the material published? Has it been updated?
• Credibility
  Is the content objective or biased? Does it cover the topic sufficiently? Is important information missing? Are the facts accurate and are references provided? Who cites the material? Is the structure clear and the language appropriate for its purpose?
• Authenticity
  Is the author who they claim to be? Is the material genuine and unaltered?

Assess whether the material is scholarly

Although subscription databases usually contain quality-reviewed articles, no source is entirely reliable. Be especially careful if you have retrieved articles from open sources such as Google Scholar, open access journals or AI search tools for scholarly material.

It can be difficult to determine whether an article is scholarly, but there are criteria that can help you in your assessment. The following links provide more information about what characterises scholarly journals and articles, and how to work with scholarly material.

• How do I know it is a scientific article?
  Learn more about what characterises scholarly journals and articles.
  
• Reading strategies – Reading scientific material
  Information on how to read and work with scholarly material.
  

Predatory journals

Predatory journals are a growing problem. As more articles are published as open access, the number of predatory journals has increased.

A predatory journal publishes articles without following established scientific quality standards. In some cases, the articles contain fabricated or inadequately reviewed research. The purpose is often to generate income from publication fees, known as article processing charges, APCs.

To attract researchers, predatory journals may use misleading practices, such as rapid or inadequate peer review, false marketing or inaccurate editorial information. Their names and appearance may resemble those of established scholarly journals, making it difficult to determine whether they are legitimate.

There is no clear definition of what characterises a predatory journal. You therefore need to evaluate articles carefully. Use general criteria to assess whether an article is scholarly. If you are unsure, consult your supervisor.

If you are a researcher, there are additional guidelines to follow. You can also seek support from the University Library.

Assessing the reliability of the material

Material included in a thesis or other scholarly work must be reviewed and evaluated. The purpose is to determine how reliable the studies are and whether the results may be inaccurate or biased. This is referred to as bias.

Bias in scholarly articles

Bias may result from how the study is designed and conducted, how participants are selected or how data are collected. A common way to assess reliability is to use a review checklist.

The Swedish Agency for Health Technology Assessment and Assessment of Social Services (SBU) provides checklists for reviewing both quantitative and qualitative studies. Many of these are designed for systematic reviews. Adapt the level of review to your stage of education and the scope of your assignment.

By critically reviewing articles, you gain a better understanding of the content and develop your critical thinking skills.

• Critical appraisal tools and checklists
  Use review checklists from SBU to support your assessment of the reliability of articles.

Bias in AI search tools

Chatbots should not be used to generate source material for a scholarly assignment. If you use a chatbot as a discussion partner, you need to be aware that its responses may be biased or contain prejudice.

This is because the language models behind chatbots are trained on material created by humans. If certain groups are underrepresented in the training data, the responses may be discriminatory, for example based on gender, ethnicity or culture. AI can then reinforce existing societal biases. The risk increases if the model is trained on AI-generated content that already contains errors. This can lead to inaccurate or biased information being spread and reinforced over time.

If you use AI search tools that retrieve sources and generate answers based on scholarly publications, you still need to be cautious. These tools often do not have access to full-text articles. Instead, their responses are based on abstracts or metadata, which may mean that important details are missing.

The quality of journals varies, and it is not always clear why certain sources are selected over others. This can result in biased or incomplete answers. Always review the sources carefully.

AI search tools can be useful in exploratory searches. However, they are not suitable for systematic searches, since it is often unclear how the search was conducted and difficult for others to reproduce it.

• What risks are associated with AI?
  Internet Knowledge, part of the Swedish Internet Foundation, describes risks related to AI, such as bias and misleading information. Available in Swedish.

How should I use sources in my work?

After completing the search process, the next step is to write and present your work. Academic writing has different requirements than many other types of writing. You must also cite all sources correctly. The links below provide resources to support you in your work.

• Copyright and plagiarism
  Information on how you may use texts and images in your work.
• Academic writing
  Guidance on academic writing and the writing process.
• Citing
  Paraphrasing, quoting and referencing styles.
• Presentation techniques
  Learn more about presentation skills.