Has AI Finally Toppled the Tower of Babel?, a post I wrote in July about how generative algorithms affect translation, grabbed the eye of my pals at the Center for European Policy Analysis (CEPA). It was published on CEPA’s website under the title “Will AI topple the European Tower of Babel” (pdf) after substantial editing and the addition of background information regarding Spain’s recent proposal that Catalan, Basque, and Galician become official languages in the European Union.
Now, Marcos Sierra from the Spanish online newspaper Voz Pópuli has utilized it in a Spanish article (pdf) where he suggests that in order to allow for the use of these three languages in the sessions of the Spanish Congress, automated algorithmic translations should be used instead of human translators.
Despite having substantially improved over the past few years, automated translations up until this point only offered results that, at best, gave a general notion of the material without much complexity and occasionally used phrases that made no sense. This was due to the fact that these algorithms initially went word-by-word through a text before learning to do so sentences at a time, then paragraphs at a time, while the more advanced ones eventually learned to employ some general context.
Translation programs have advanced significantly in recent years with the introduction of generative algorithms and the use of vast computing power in their training, and those that are now being utilized by the general public actually improve every day. A lot of us have experimented with generative We used algorithms like ChatGPT or Claude to edit and translate our own texts into different languages, and we were amazed by the quality of the output: there were no odd words or questionable grammar. For some manuscripts, a skilled human translation may still be advised, but most of the time, an algorithm is adequate.
Additionally, if given writings that we have written, these algorithms can mimic our writing style to the point where even when translated into another language, the texts still read as though they were written by us.
What’s the next step for us? First of all, it’s likely that many translators are assessing their future and the extent to which they can contribute to a text. Someone who is employed as a simultaneous interpreter, such as those employed by…
