Examples
Worked examples
- Is an instance
Using an AI translation tool to produce a first draft of a non-English manuscript's English version, which is then edited and verified by a bilingual co-author
- Is an instance
Machine-translating a validated survey instrument and then independently back-translating it to check the two versions still ask the same question
Counter-examples
Looks similar, but isn't
- Not an instance
A certified human translator producing a translation with no AI tool involved is standard professional translation, not AI translation
- Not an instance
AI condensing text within one language is ai-summarisation, not translation
Editorial commentary
AI translation ranges from routine neural machine translation (the kind built into many browsers and word processors) to LLM-based translation capable of handling idiom, register, and technical terminology with more sensitivity than older statistical systems — but not with guaranteed accuracy, particularly for domain-specific or legally sensitive text.
Where the stakes are highest
For most manuscript-writing use, AI translation is treated similarly to other AI-assisted language editing: broadly permitted with disclosure norms varying by publisher, since the underlying research content and argument originate with the human author. The stakes rise sharply for two categories of document: informed consent forms and validated research instruments (surveys, clinical outcome measures). A literal translation of a consent form can preserve grammatical correctness while losing the specific legal and ethical meaning the original language was carefully drafted to convey — IRBs/RECs commonly require a documented back-translation and reconciliation step for consent materials regardless of whether the initial translation was AI-assisted or human, precisely because meaning equivalence, not just linguistic correctness, is what’s being certified. See informed consent in research.
How this differs from related AI-band terms
- vs. AI summarisation: translation aims to preserve the full original meaning across a language boundary; summarisation deliberately discards detail to shorten text within one language.
- vs. AI in qualitative coding: a distinct task — translating raw non-English qualitative data before coding is a preparatory step, not the coding itself, and errors introduced at the translation stage can propagate into every downstream code.
Practical guidance
Treat AI-translated text the same as any AI-assisted output: verify it against the source (ideally via independent back-translation for consent and instrument materials), disclose substantive AI translation use per the target journal or funder’s policy, and never rely on AI translation alone for legally operative or safety-critical participant-facing text.
Also known as
Machine translation · Neural machine translation
Machine-readable encodings
Use in your systems
<role vocab="credit"
vocab-identifier="https://casrai.org/dictionary/"
vocab-term="AI translation"
vocab-term-identifier="https://casrai.org/dictionary/term/ai-translation" />{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://casrai.org/dictionary/term/ai-translation",
"name": "AI translation",
"identifier": "https://casrai.org/dictionary/term/ai-translation",
"description": "The use of an AI system -- most commonly a neural machine translation model or a general-purpose LLM used for translation -- to convert text from one natural language into another, applied to manuscripts, participant-facing materials, survey instruments, or informed consent documents.",
"inDefinedTermSet": "https://casrai.org/dictionary/domain/genai-disclosure#set",
"url": "https://casrai.org/dictionary/term/ai-translation",
"sameAs": [
"Machine translation",
"Neural machine translation"
],
"license": "https://creativecommons.org/licenses/by/4.0/",
"publisher": {
"@id": "https://casrai.org/#organization"
},
"author": {
"@id": "https://casrai.org/#editorial-team"
},
"datePublished": "2026-05-21T01:56:47",
"dateModified": "2026-08-22T14:51:53",
"inLanguage": "en-GB",
"isAccessibleForFree": true
}
