AI tools for research

AI research tools can help you find relevant papers faster, get a first idea of what they say, extract key data into tables and see whether later studies supported or contradicted a finding. They are most useful in the exploratory phase of a literature search. They do not replace reading the sources yourself, and whether you may use them in assessed work depends on your institution’s rules.

Note: your examination regulations, your supervisor or, for papers, the journal’s policy decide whether and how AI may be used. Many institutions require you to declare AI use, for example in a statement at the end of the thesis. Use AI as an aid for searching and understanding literature, never as the author of your text.

What AI research tools can do

• Semantic search: you type a question in plain language and get papers that match the meaning, not just the keywords.
• Summaries: short summaries of abstracts or full texts to help you decide what to read.
• Data extraction: tables with methods, sample sizes or outcomes across many papers, useful for a literature review matrix.
• Citation context: information on how later papers cite a study, for example whether they support or dispute it.
• Discovery: suggestions of related work based on papers you already know.

What they cannot do

• Guarantee completeness. Coverage depends on which databases the tool indexes, and some publishers’ full texts are not included.
• Judge quality reliably. A summary does not tell you whether a study’s design supports its conclusions.
• Replace a documented search strategy. For systematic work you still need reproducible searches in subject databases.
• Take responsibility. Errors in a summary become your errors if you copy them.

Three tools compared

	Elicit	Scite	Consensus
Main purpose	Find papers and extract data from them	Show how papers cite each other	Answer research questions from papers
Strength	Tables across many studies, screening support	Citation statements classified, e.g. supporting or contrasting	Quick overview of findings on a question
Good for	Structuring a literature review	Checking whether a key finding has held up	Getting oriented in an unfamiliar topic
Watch out for	Extraction errors in tables	Classification is automated and can be wrong	Simplified answers to complex questions

Status October 2026, based on provider documentation. Features, coverage and plans change frequently.

Elicit

Elicit is built around searching for papers and pulling information out of them. You can ask a research question, get a list of relevant papers and add columns such as method, population or main outcome. This is helpful when you need to compare many studies quickly, but every extracted value should be checked against the paper.

Scite

Scite focuses on citation context. Instead of only counting citations, it shows the sentences in which other papers cite a study and classifies them, for example as supporting or contrasting. This helps you see whether a frequently cited result has been confirmed or challenged.

Consensus

Consensus works like a search engine that answers questions with findings from research papers and links to the sources. It is useful for orientation, but complex or contested questions rarely have a one-line answer, so read the underlying studies.

Other tools you may meet

Citation-network tools such as Connected Papers, ResearchRabbit or Litmaps visualise how papers relate to each other. Semantic Scholar is a free academic search engine with AI features. General chatbots can help explain a concept, but they should not be used as a source of references unless they cite verifiable papers.

A responsible workflow

1. Clarify the rules with your examination regulations and supervisor before using AI.
2. Start with a conventional search in subject databases, following our guide to literature search.
3. Use AI tools to explore: find additional papers, check citation context, build a first comparison table.
4. Verify every paper in a database or library catalogue and read the relevant parts yourself.
5. Save sources in a reference manager, not just in the AI tool.
6. Document your use of AI tools as required, including which tool, when and for what.

Data protection

Before uploading unpublished manuscripts, interview data or confidential material, check where the provider processes data and whether uploads are used for training. Sensitive research data usually should not be uploaded at all.

Related tools

AI tools that help with language and style are covered in academic writing assistants. How plagiarism and AI detection work is explained in plagiarism checkers. For the publisher side, including policies on AI in reviewing, see peer review. All comparisons are in the tools section.