AI tools have genuinely changed how efficient research can be — and generated enough hype to bury the honest question: which tools actually help, for what, and where do they fail? Here's our working map, tool by tool, with the same standard throughout: AI assists your thinking; it never produces the substance you submit. (The full ethical framework is in using AI ethically in research.)
Literature discovery and mapping
- [Elicit](/learn/elicit-for-literature-review) — asks research questions across the academic literature and extracts findings into comparison tables. The strongest screening accelerator for reviews; every extraction still needs verification against the paper.
- [Research Rabbit and Connected Papers](/learn/research-rabbit-vs-connected-papers) — citation-network mapping from seed papers. The fastest cure for 'have I missed a key paper?' anxiety, and free.
- [Scite and Consensus](/learn/scite-vs-consensus) — citation-context and evidence-synthesis engines: Scite shows whether later papers support or contrast a claim; Consensus summarises what the literature says about a question.
Working with your own sources
[NotebookLM](/learn/notebooklm-for-research) (Google) is the standout here: upload your PDFs and it answers questions grounded only in those sources, with clickable citations — far less hallucination-prone than open-ended chatbots, and genuinely useful for organising a literature review corpus.
General assistants: ChatGPT, Claude, Gemini, Perplexity
The general LLMs are best treated as tireless, occasionally overconfident research assistants: superb for brainstorming, structuring, explaining methods, critiquing your drafts and helping with analysis code; unreliable for facts and citations, which they still fabricate confidently. Perplexity partially fixes the citation problem by searching and citing sources — see Perplexity vs ChatGPT for research for when each fits. None of them may write your thesis: that line is misconduct, and it's also where our answer on ChatGPT and theses begins.
The workflow that actually works
- 1Map the field: Research Rabbit / Connected Papers from 2–3 seed papers.
- 2Screen at scale: Elicit for questions and extraction tables; Consensus for quick evidence orientation.
- 3Verify claims you'll build on: Scite's citation contexts + reading the actual papers.
- 4Organise and interrogate your corpus: NotebookLM over your PDF library.
- 5Think and refine with a general LLM — outlines, critique, code — while every fact, citation and sentence you submit remains verified and yours.
(1) Anything AI-generated is a draft input to your judgement, never submitted text. (2) Every citation gets verified in the original source — AI tools, including the good ones, mis-extract and fabricate. Follow both and these tools compress months of work; skip them and the tools compress your credibility instead.
Want a mentor to set up an ethical AI workflow for your specific thesis? That's part of our AI-for-researchers mentoring — and our free tools handle the classic chores (citations, sample size, timelines) without any login.
Frequently asked
Which AI tool is best for literature review?+
For discovery: Research Rabbit or Connected Papers. For screening and extraction at scale: Elicit. For organising and querying papers you've collected: NotebookLM. Most strong workflows combine all three stages.
Are these AI tools free?+
Research Rabbit is free; Connected Papers, Elicit, Scite, Consensus and Perplexity run freemium models with usable free tiers; NotebookLM is free with a Google account (limits apply). Capabilities and pricing change quickly — check current terms.
Will using AI tools get me in trouble with my university?+
Assistive use — discovery, organisation, feedback on your own writing — is broadly accepted and increasingly expected. Trouble comes from submitting AI-generated text or unverified citations as your own. Disclose per your university's policy and keep the substance yours.
phdguide's mentors are senior academics, former supervisors, statisticians and publication specialists with 25+ years of combined experience guiding MBA, MPhil and PhD scholars from topic to viva.
Ethical, compliant guidance: We provide academic support, mentoring, analysis, editing and structuring — not authorship. Your work stays compliant with university policies.