The literature review is often the longest chapter to write and the one scholars rewrite most. A new generation of AI-powered research tools can help you discover papers, see how studies connect and organise your reading. Used well, they save weeks. Used carelessly, they produce shallow reviews with missing or invented sources.

Types of AI research tools

  • Semantic search tools find papers by meaning rather than exact keywords
  • Citation-mapping tools show networks of related papers visually
  • Summarising and extraction tools pull key findings or methods from papers
  • Reference managers such as Zotero and Mendeley organise and cite sources

A responsible seven-step workflow

  1. Define your scope. Write your research questions and inclusion criteria before you search.
  2. Search the core databases first. Use Scopus, Web of Science, PubMed or discipline databases with a documented search string. AI tools supplement this — they do not replace it.
  3. Expand with citation mapping. Use AI tools to find highly connected papers you may have missed.
  4. Screen titles and abstracts yourself against your criteria, and record why papers were excluded.
  5. Read the key papers in full. Summaries can miss limitations, context and nuance.
  6. Build a synthesis matrix — a table of author, year, context, method, findings and gaps.
  7. Write by theme, not by paper. Compare and contrast studies to build towards your research gap.

For systematic reviews: follow PRISMA

If you are conducting a systematic review or meta-analysis, report your process using the PRISMA 2020 statement and flow diagram. Record exactly which databases and tools you used, including any AI tools used for screening, so your review is reproducible.

Pitfalls to avoid

  • Unverified citations — check every reference in the original source
  • Over-reliance on abstracts — they rarely show limitations
  • Descriptive reviews that summarise papers one by one without synthesis
  • Uploading paywalled PDFs to tools in ways that breach publisher licences

The part only you can do

A strong literature review shows your critical judgement: which studies are robust, where findings conflict, and what gap your study fills. That argument is what examiners assess — and it is the part AI cannot write for you.