Background Research & Evidence Synthesis

AI tools can help streamline the early stages of research by finding relevant papers, summarising academic content, and even pointing to gaps in the existing literature. They're particularly useful when you're getting familiar with a new topic, developing your research question, or exploring unfamiliar areas. While these tools don't replace a rigorous, systematic literature review, they can save time and support your background research.

Background Research

What AI can help with

AI can help you:

• Find relevant papers faster by looking at meaning, not just keywords.
• Summarise studies and highlight key findings, limitations, and methods.
• Identify gaps in the literature by comparing many papers at once.
• Discover related work and visualise how studies are connected.
• Organise your reading with tools that can "remember" and search through the documents you've saved.

Different tools have different strengths, so it's worth trying a few. Always double-check any AI-generated summaries by reviewing the original papers directly.

Tools

Hear from researchers using them

Walkthroughs and write-ups from people using these tools in practice:

• This AI Synthesizes 500 Papers So You Don't Have To — a video walkthrough of Elicit.
• Scite AI Just Changed How I Trust Research — a video tutorial on Scite.
• Best AI Tools for Literature Review — ResearchRabbit's own round-up.

---

Evidence Synthesis

AI is now used at several points in the review workflow, but we do not recommend running a review end to end on tools like Undermind or Elicit. Where the work has to be rigorous, comprehensive and repeatable, established methods still set the standard — use AI to support manual methods and expert judgement, not to replace them.

Three resources are worth knowing before you start.

1. Joint position statement on AI in evidence synthesis (2025)

Four shared expectations from Cochrane, the Campbell Collaboration, JBI and the Collaboration for Environmental Evidence — see the full statement:

• Stay accountable. You remain responsible for the review, whatever tools you used — for choosing them, for how you applied them, and for their effect on your methods and conclusions.
• Justify the choice. Show the tool suits your specific review and does not weaken its reliability.
• Declare it. Wherever AI makes or suggests a judgement, report the tool, version, dates, prompts and validation steps under PRISMA/ROSES.
• Follow the usual rules. Copyright, licensing, confidentiality, privacy and data-protection duties are unchanged by using AI.

2. RAISE — responsible use of AI in evidence synthesis

RAISE turns those expectations into a working method: how to pick a tool for each task, pilot it on your own material before trusting it, record what you did in your protocol and methods, and handle the legal and ethical risk. It is designed to sit alongside existing guidance — the Cochrane Handbook, Campbell and JBI methods, PRISMA and ROSES — rather than replace it.

Systematic Reviews has a short recorded introduction covering both the position statement and RAISE.

3. Repositories of AI tools for evidence review and synthesis

Two places to check before committing to a tool:

• MetaEvidenceAI — a living registry of published evaluations of AI tools for evidence synthesis, led by Dr Ciara Keenan. An existing evaluation is the fastest support for your justification, and you can add your own.
• King's College London — AI Tools in Evidence Synthesis — an overview of tools for building search strategies, finding relevant articles, and screening, extracting and synthesising data.

Putting this into practice

Use all three to shape your own policy, training and day-to-day practice, so that a reader of your review can see what AI did and why that was a defensible choice.