As with so many things in life, the truth lies somewhere in the middle.
It is generally acknowledged that when it comes to languages, AI technology has been gaining momentum fast since the (in)famous ChatGPT first burst on to the scene. Hand in hand with this breakthrough in technology, there’s now a growing opinion doing the rounds that computers are capable of just about anything in the field of language – and can do it even better than humans. But, in fact, nothing could be further from the truth!
Admittedly, things have come a long way since the often laughable and even mind-bending “google translate” translations that we have all come across at some stage before.
Nevertheless, even using the best AI tools available, the results produced by machine translations still tend to remain mediocre at best in terms of style and creativity. Worse still, they can be highly unreliable and misleading when it comes to accuracy (this sort of, often costly, machine-generated nonsense mainly includes context-sensitive interpretations, subject-specific terminology and individual nuances that may be totally misunderstood).
For large volumes, or technical texts where style may be deemed less important, whereas AI systems may provide significant time-savings for producing ‘preparatory’ work, they never (!) offer benefits in terms of quality. This means that human input in the form of proofreading and editing is still required to set the standard for the quality of the final result.
So, to sum up, machine translations should only be considered as an option for generating large volumes (especially when combined with deadlines that are too short) and when budget is the absolute priority. But in that case, an important choice needs to be made between being content to carry out a superficial and purely linguistic revision of the text, or conducting an in-depth and substantive (comparative) review that focuses on quality and accuracy.
Making that choice will tend to be the deciding factor for establishing the total cost of the job and the level of quality linked to it:
1) with a simple revision, only the machine translation itself (in the target language) is looked at. This means that only pure language-related mistakes (spelling, grammar) can be identified and corrected and – best-case scenario – the most glaring errors and mistranslations detected. Of course, taking this option is a great deal less time-consuming (i.e. cheaper). But because the source text is not carefully looked at, any genuine (substantive) translation errors will slip through the net – potentially to the embarrassment and harm of the client. This means that a simple revision does not actually guarantee the accuracy of the translations. Admittedly, sentences produced by AI will usually read fluently enough, but they may also differ substantially in content from what is written or intended in the original text. Unfortunately, however, this is the method offered by some translation agencies (often without specifying that only a “simple” revision has been carried out). And the aim of this, of course, is to be able to provide “rock-bottom” prices and maximise profits.
2) the way we see it at Dynamics, a full comparative review is the only option for more or less matching the quality of a human translation and guaranteeing that the final result is correct in terms of substance. But it goes without saying that a thorough review such as this, in which both languages are compared, sentence by sentence, by an expert native speaker, requires much more time (and knowledge). And hence the ultimate cost of the job is likely to be on a par with a 100% human translation anyway.
Whatever your situation happens to be or the choice you make, we’ll be happy to assist you. If you already have a machine translation and would like it reviewed by one of our specialists, we’re ready to give you an objective assessment of its quality – plus we’ll always offer you the best terms for reviewing all your texts, based on your own criteria and in line with the preferences you set.
