Λογοτεχνική μετάφραση και Τεχνητή Νοημοσύνη

17 March 2024 - Sunday

AI and literary translation (cont'd) | Prix Jean d'Ormesson longlist

       AI and literary translation (cont'd)

       With yesterday's posts on AI and translation and the posting of The Club of True Creators-review, I really should have put two and two together, since the latter more than touches upon the subject of the London Book Fair panel on AI and Literary Translation and the two articles I linked to, as the publisher of The Club of True Creators, the new Rossum Press, have explicitly embraced a publishing- and translation-model based on Artificial Intelligence.
       As they explain/maintain:

Using a system of AI-assisted team translation, our skilled editors are able to create high quality literary translations with a fraction of the resources which traditional methods require.

Every word of the AI-generated draft translation is carefully weighed by a professional stylist of the target language, and we work closely with our authors at every step along the way.

       Like it or not -- and many people (not just, but especially translators) really, really don't like it --, this is (at least a significant part of) the future, especially for popular and genre fiction (and, for example, manga), and, if nothing else, props to Rossum Press for making it very clear that this is how they operate. (Well, they might have mentioned it in the translator-creditless book itself as well .....)
       The 'machine translation + (human) editing' model seems likely to become the dominant one -- with the amount of editing varying widely (as it does already: one should never overlook that a lot of entirely human translations are terrible, not least because they are often published without much editorial oversight or involvement)).
       One of the reasons given for so little being published in translation is the cost involved. The use of machine-translation -- to whatever extent -- can reduce those costs drastically -- but will the final product justify those (reduced) costs, or are we possibly losing too much ?
       (I do note -- and I do think this isn't acknowledged nearly enough -- that, both historically and currently, a lot of (human) translation of literary work, both popular and 'literary', is really bad. (Admittedly, the main reason for this -- (many) publishers simply don't care about the (end-)product-- applies to any form of translation, i.e. won't be rectified by greater reliance on machine-aided translation.))

(Posted by: M.A.Orthofer)    - permanent link -