10 AI Tools Worth Knowing for Students and Researchers

A literature review used to start with a database and a long afternoon. It now often starts with a question typed into an AI tool. That is faster, and it is also where the new mistakes come from: a summary that sounds right, a citation that does not exist, a paper described as saying the opposite of what it says.

This list covers ten tools that students and researchers actually meet. For each we say what the vendor says it is for, one honest limit, and a note on using it with integrity. It is an editorial list, not a ranking, and not a test.

Which tool does which job?

Which tools help me find papers?

1. Semantic Scholar

What it is for: Semantic Scholar describes itself as a no-cost, AI-powered search tool for scientific papers, and it also offers open datasets and APIs for developers. It is run by Ai2, a nonprofit research institute. The limit: a good search tool shortens the hunt, but you still have to read the paper and judge it. Coverage and quality of the ranking can differ between fields, so run the same query somewhere else too. Integrity note: a result list is not a bibliography. Cite what you have read.

2. Consensus

What it is for: Consensus says it helps people find, sort and examine peer-reviewed studies, with AI-written answers, study summaries and a 'consensus meter' for yes/no questions. It is also in our research tools shortlist. The limit: the summaries are AI-generated and can contain errors, so follow the links to the papers. A meter that appears to show agreement is a convenience, not a systematic review. Integrity note: if your question is a clinical or high-stakes one, treat the output as a starting point for reading, never a conclusion.

3. Elicit

What it is for: Elicit describes itself as an AI research assistant that searches academic papers and produces citation-backed reports, with tools for systematic reviews and for organising sources. The limit: an automated report inherits any gap in the papers it finds. If a relevant study is missing from its search, it is missing from your review. Integrity note: for a formal systematic review, follow your field's reporting standard and record how you used the tool, because a reviewer may ask.

4. Perplexity

What it is for: Perplexity's help centre describes an AI-powered search engine that searches the web in real time and shows the sources behind each answer, with links so you can check them. We cover it in our Perplexity Pro listing. The limit: it searches the open web, not only scholarship, so a cited source might be a blog or a press release. Judge each source as you would any other. Integrity note: a linked source proves the model had something to cite, not that it summarised it correctly. Open the link.

Which tools help me map a field?

5. ResearchRabbit

What it is for: ResearchRabbit says it helps you move from a few papers you know to a wider view of your field, analysing the papers you pick and suggesting related studies, authors and topics. The limit: it expands from your starting point, so a biased or narrow seed list produces a narrow map. Integrity note: record your search method. 'I followed suggestions from a tool' is a method you should be able to describe.

6. Connected Papers

What it is for: Connected Papers presents itself as a service for finding and exploring academic papers. We could confirm only that general description from its pages, so check its site for how the exploration works. The limit: relatedness is not quality. A paper that sits at the centre of a map may be widely cited because it is wrong or contested. Integrity note: use maps to find reading, then judge each paper on its methods and evidence.

Which tool helps me check a paper's standing?

7. scite

What it is for: scite says its 'Smart Citations' show whether later studies have supported, challenged or simply discussed a given paper or claim. That is a different question from a raw citation count. The limit: deciding whether a sentence 'supports' or 'disputes' a paper is a judgement, and an automated classifier can get it wrong. Integrity note: if a paper's standing matters to your argument, read the citing passages yourself before relying on the label.

Which tools help me organise and write?

8. Zotero

What it is for: Zotero describes itself as a research assistant that helps you gather, sort, annotate, cite and share sources, capturing them as you browse and building citations and bibliographies inside word processors. The limit: it organises sources; it does not tell you which are reliable, and a tidy library can hide a weak reading list. Integrity note: it is often the tool that protects you from a bad citation, because it records where you actually got a source. Keep it honest by adding only what you have opened.

9. NotebookLM

What it is for: Google describes it on its product page as an AI research tool and thinking partner. The limit: this is our thinnest entry. Google's page, which now carries the title 'Gemini Notebook', did not state what it works from or how it handles sources when we read it, and its help article could not be fetched. Read Google's current documentation before relying on it. Integrity note: whatever any tool says about your documents, check it against the documents.

10. Claude

What it is for: Anthropic describes Claude as an AI collaborator for multi-step work: drafting and editing documents, analysing data and working from files you provide, with editable output you review. It is also in our products directory. The limit: it is a general assistant, not a scholarly database, so it may state a fact or reference with confidence that it cannot back up. Give it your own material to work from and verify the rest. Integrity note: if you let any assistant draft prose, you remain responsible for every claim in it, and your institution may require you to disclose the use.

How could I put these together without fooling myself?

Our guides on how to evaluate AI models and reading benchmark claims critically apply the same habit: ask who made a claim, what it rests on and whether it can be checked.

What about academic integrity?

Rules about AI use differ between institutions, courses, funders and journals, and they change. Some allow AI for brainstorming and forbid it for drafting; some require disclosure; some forbid it entirely. We cannot tell you what yours says. Find the current policy, read the assignment or submission guidance, and ask when it is unclear. When in doubt, disclose.

What are the limitations of this list?

We read each tool's own description and nothing more. We did not run the tools, compare their results or measure accuracy, and vendors' descriptions are vendors' claims. Features, availability and terms change, so check each tool's current page. The list leaves out many other tools, and our omission of ChatGPT reflects only that we could not verify its official description on the day we checked. See our AI research tools category for more, and our corrections page if you spot an error.