What is Machine Translation (MT)?

Machine translation is translation produced by software rather than a human. Modern systems are neural: dedicated engines such as DeepL and Google Translate, and increasingly large language models, which can follow instructions about tone, terminology, and placeholders in a way classic MT engines cannot.

MT quality varies with language pair, domain, and context. Short UI strings are actually among the harder cases, because a single word like "Post" or "Book" is ambiguous without knowing where it appears. Supplying context, surrounding keys, screenshots, or a description of the screen, measurably helps any engine, and matters even more for LLMs.

The common production pattern is machine translation post-editing (MTPE): MT produces a first pass, a human reviews and corrects it. For low-stakes or internal content, raw MT may be acceptable; for marketing and legal copy, human review or full human translation remains the norm. LangSyncer integrates AI translation directly into the editing workflow so the first pass is one click.