Artificial intelligence and translation: what does the future hold?
Let’s be honest! Quite frankly, here at Translation & More, we are not great fans of machine translation (MT) carried out with the help of AI; our mantra has always been the craftsmanship of the work (so-called ‘artisanal intelligence’, as the well-known entrepreneur Diego Della Valle put it), which has little in common with the sterile product of a machine.
However, we understand that if you need a translation for internal business use to be produced quickly, it can sometimes be useful to get an understandable text in just a few seconds.
Of course, in order to transform this draft into an externally marketable or publishable text, a professional ‘human’ translator can make all the difference by carrying out what is called ‘post-editing’, i.e. the complete revision of the translation obtained with AI, with the correction of grammatical, lexical and morphosyntactic errors. The human translator will also detect any inconsistencies in the translation, standardising the language used and creating an appropriate style.
Permit us, however, to emphasise a couple of things. It is well known that the translation process is (or was??) highly creative and involves a thorough analysis of the text, whatever the subject matter. No one ‘gets under the skin’ of your text like the translator, often detecting any contradictions or inaccuracies. A corporate presentation, for example, is a business card, the ‘Sunday best’ with which one presents oneself to the world, indeed to a multiplicity of worlds with different cultural backgrounds requiring an appropriate sensitivity.
A high degree of accuracy is also required in the translation of contracts, legal deeds, and documents intended for Courts and Tribunals, where machine translation often gets lost in the mindless prose of ‘legalese’. There is therefore an urgent need for very careful monitoring and a good knowledge of the subject.
The same difficulty is encountered with highly specialised technical texts, complex texts requiring in-depth syntactic reworking and, of course, with literary texts. Machine translations of the latter are depressing, or perhaps on the other hand, comforting, because they show that human ingenuity has not yet been overwhelmed and replaced…
Should we therefore demonise Machine Translation? Not necessarily, if we know how to use it properly. As already mentioned, machine translation can be used to get a quick understanding or an ‘off-the-shelf‘ text that is of immediate use.
If the purpose is different, it is best to seek the support of a professional translator, with whom you can establish a relationship of trust. Whether it is the post-editing of your text or a fully human translation, the professional will build a customised project for you, creating a dedicated glossary, using an appropriate style and, what is most important, taking responsibility for his own work. Translators are often called upon to provide a sworn translation, i.e. they take an oath in which they attest to the translation’s conformity with the original.
Another major issue is that of data protection: if data is fed into artificial intelligence, especially if it is open source, in order to instruct the system, it will be usable by anyone, which could cause serious harm to patent owners and pose a danger for company secrets.
In the end, there is plenty to think about when deciding which choice to make, don’t you think?
