Examples
Worked examples
- Is an instance
Using Elicit to generate a first-pass list of candidate studies for a scoping review, then manually screening each against inclusion criteria
- Is an instance
Asking an LLM to suggest key papers on a topic and independently verifying every citation exists and says what the tool claims before using it
Counter-examples
Looks similar, but isn't
- Not an instance
Running a Boolean search in PubMed or Scopus with no AI ranking or summarisation involved is standard database search, not AI-assisted literature search
- Not an instance
Pasting an already-retrieved paper into a chatbot and asking for a summary is ai-summarisation, not literature search
Editorial commentary
AI in literature search covers tools that sit upstream of reading — they help a researcher find relevant work, not condense work already found. This includes semantic-search and recommendation engines (Semantic Scholar, Scite, Consensus), LLM-based research assistants that synthesise across retrieved papers (Elicit, Undermind), and the ad hoc practice of asking a general-purpose chatbot to name relevant literature on a topic.
The specific risk profile
Two failure modes are distinct from ordinary database search and from other AI use cases on this site. First, hallucinated citations: an LLM asked to recall or synthesise literature from parametric memory (rather than a grounded retrieval index) can generate plausible-looking but nonexistent papers, or attribute a real paper’s findings incorrectly. Every AI-suggested citation needs independent verification against the actual source before it is used or cited — see AI output verification. Second, recall and coverage bias: AI ranking systems trained predominantly on well-cited, English-language, open-access literature can systematically under-surface older, non-English, or low-citation work, which matters for any review claiming comprehensiveness.
How this differs from related AI-band terms
- vs. AI summarisation: literature search is a discovery task (finding candidate sources); summarisation operates on sources already identified. A workflow commonly uses both in sequence.
- vs. AI in qualitative coding: a different research stage entirely — coding applies to a study’s own primary data (interviews, open-ended responses), not to the secondary literature.
Reproducibility and disclosure
Unlike a documented database query, AI-tool retrieval results are frequently non-deterministic and can change between runs as the tool’s underlying index or model updates, which complicates the kind of reproducible, auditable search strategy that a systematic or scoping review protocol (e.g. PRISMA) expects. Best practice is to record the tool name, version, and date of use, and to treat AI-assisted discovery as a supplement to, not a replacement for, a documented database search strategy — consistent with ICMJE and COPE’s general position that AI-assisted research steps belong in the Methods section with enough detail to be understood and, where possible, repeated.
FAQ
Does using an AI literature tool need to be disclosed in a manuscript? If the tool materially shaped which literature was included or how it was characterised, yes — treat it the same as any other AI-assisted research step under current journal disclosure norms; see generative AI disclosure statement.
Also known as
AI-powered literature search · LLM literature review
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="AI in literature search"
vocab-term-identifier="https://casrai.org/dictionary/term/ai-in-literature-search" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/ai-in-literature-search",
"name": "AI in literature search",
"identifier": "https://casrai.org/dictionary/term/ai-in-literature-search",
"description": "The use of AI-powered discovery and retrieval tools (e.g. Elicit, Consensus, Scite, Undermind, Semantic Scholar's semantic-search features, or a general-purpose LLM queried directly) to identify, rank, filter, or synthesise relevant scholarly literature, as distinct from using AI to condense text a researcher has already found.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/genai-disclosure#set",
"url": "https://casrai.org/dictionary/term/ai-in-literature-search",
"alternateName": [
"AI-powered literature search",
"LLM literature review"
],
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"author": {
"@id": "https://casrai.org/#editorial-team"
},
"datePublished": "2026-05-21T01:57:14",
"dateModified": "2026-08-22T14:51:46",
"inLanguage": "en-GB",
"isAccessibleForFree": true
}
