I recently read a Guardian article about AI and translators in Europe. The article moved between literary translation, machine translation, post-editing, technical translation, and the broader language services industry as if they were all variations of the same work.

That is exactly the problem.

"Translation" is no longer a useful word when it is used this loosely.

Literary translation, children's books, technical documentation, SaaS UI, marketing copy, support content, and post-editing machine output do not operate under the same rules.

When preserving the surface is the work

In literary translation, preserving detail is often the work. The translator may need to protect ambiguity, rhythm, silence, repetition, or the relationship between text and image.

I once saw a Japanese translation of a Brazilian children's book that made this very clear. On one page, the illustration showed a rainbow. The Portuguese text referred to the colors with a loose word meaning something like "the group" or "the gang."

In context, yes, it was the rainbow. But the original text did not say "rainbow." The Japanese translation did.

That may sound like a small clarification. But it changes the page. The original left that discovery to happen between the illustration, the adult reading aloud, and the child looking at the page. Making the intention explicit may sound helpful. In that context, it changes the work.

When preserving the surface breaks the work

In SaaS localization, the opposite is often true. Preserving the surface of the source text can break the user experience. A CTA, onboarding step, support message, or error message may need to be rebuilt around intent, user context, and expected action.

In one context, extracting intent can destroy the work. In another, failing to extract intent can destroy the user experience.

Borrowing prestige from a different kind of work

This is why I find many "AI vs human translators" discussions misleading. The issue is not whether AI can "translate." The issue is that people are using examples from one kind of language work to defend another.

When commercial translation vendors point to literary translation as proof that the entire industry still needs the same human role, they are not clarifying the debate. They are borrowing prestige from a different kind of work.

If we keep treating all of this as one profession called "translation," we will keep having the wrong conversation.