AI translation speed is forcing translators to rethink what their work is worth

Machine translation is changing what translation work is worth, not just how long it takes. French literary translator Yoann Gentric tested DeepL in 2022 and found it handled sentence nuance clumsily. When he tried again a few years later, he told The Guardian that quality had improved significantly. Berlin-based German-English translator Laura Radosh said her requests from universities, professors and museums fell from about four a month to about one the following year. Some remaining work shifted to post-editing, which she found is sometimes no faster than translating from scratch. A joint survey of French authors' associations, cited by The Guardian, found 79% of translators believe AI threatens to replace all or part of their work. A 2025 UK survey reported that 84% of translators predicted falling demand for human translation. The Guardian also reported that one German translator's technical rate dropped from 0.80 euros to 0.60 euros per line. The article argues the profession is splitting between cheap machine tasks and human-accountable work such as literary, legal and diplomatic translation. It is worth reading for its pricing argument: AI may not replace translators outright, but it can shrink what clients will pay.