Examples
Worked examples
- Is an instance
Using an LLM to produce a first-draft abstract of a completed paper, which the authors then rewrite and verify against the full text
- Is an instance
Summarising a batch of already-retrieved papers' abstracts to triage which merit full-text reading, with every summary spot-checked against the source
Counter-examples
Looks similar, but isn't
- Not an instance
Discovering which papers to read in the first place is ai-in-literature-search, a distinct upstream task
- Not an instance
Producing a shorter version of text by simple truncation or keyword extraction with no learned model is not AI summarisation
Editorial commentary
AI summarisation is one of the most common assistive uses of generative AI in research work, and one of the easiest to use carelessly, because a fluent summary reads as trustworthy regardless of whether it is accurate.
The specific risk: what condensation tends to lose
Summarisation systematically drops caveats, effect-size context, sample limitations, and conflicting evidence more readily than it drops the headline claim, because those are exactly the details a compression objective treats as low-priority. A summary can therefore overstate certainty relative to the source even when every individual sentence is technically accurate. Summaries can also silently misattribute a finding to the wrong source when condensing across multiple documents. Independent verification against the source text — see AI output verification — is the mitigation every publisher and funder disclosure policy assumes is happening before an AI summary is used or cited.
How this differs from related AI-band terms
- vs. AI translation: translation aims to preserve full meaning across a language boundary; summarisation deliberately discards detail within the same language to produce something shorter.
- vs. AI in literature search: search finds candidate sources; summarisation operates on sources already identified, often as the next step in the same workflow.
Disclosure
Where an AI-generated summary appears in or materially shaped a manuscript — an abstract draft, a synthesis of related work, a plain-language summary — current publisher guidance (ICMJE, JAMA, SAGE) treats this the same as other substantive AI-assisted writing: disclose the tool and its role in Methods or Acknowledgements, and retain full author responsibility for accuracy. Routine, lightly-assistive use (e.g. asking for a tighter phrasing of a sentence the author already wrote) is generally exempted by most policies; generating novel summarising content is not.
Also known as
Automatic summarisation · LLM summarisation
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="AI summarisation"
vocab-term-identifier="https://casrai.org/dictionary/term/ai-summarisation" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/ai-summarisation",
"name": "AI summarisation",
"identifier": "https://casrai.org/dictionary/term/ai-summarisation",
"description": "The use of a generative AI system to condense a longer text -- a paper, dataset documentation, meeting notes, or a set of retrieved sources -- into a shorter form, typically preserving what the tool judges to be the key points while omitting detail, qualification, and nuance.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/genai-disclosure#set",
"url": "https://casrai.org/dictionary/term/ai-summarisation",
"alternateName": [
"Automatic summarisation",
"LLM summarisation"
],
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"author": {
"@id": "https://casrai.org/#editorial-team"
},
"datePublished": "2026-05-21T01:56:53",
"dateModified": "2026-09-05T14:21:51",
"inLanguage": "en-GB",
"isAccessibleForFree": true
}
