Maybe I’m not super up to date on AI stuff, but I worked as a translator for a year, and AI (they used ChatGPT and DeepL) still made a bunch of mistakes that you’ll immediately notice when you speak the language. It feels like their training input had a bunch of older, Google-translated articles in them that were just bad. Maybe an AI trained specifically for translation with curated learning material and a “teacher” who corrects mistakes can get closer to replacing human translators, but it’d still miss the cultural context of certain words and phrases that are in a translator’s passive vocabulary, at least in less widely spread languages.
That being said, it’s definitely harder to make a career out of translating because companies who don’t know any better just use AI instead. As long as they get their point across (and make money), they don’t care about the finer details.
Sure, a skilled human is still better at the job. But you don’t always need to capture every nuance. And AI does it at the fraction of the cost.
I see this with lots of German product descriptions on big store fronts like Amazon. They often seem entirely machine translated. It’s not great, but “good enough” and serviceable.
Machine translation can also increasingly shifts the process from the sender of the message to the recipient. It used to be that the web page of a Vietnamese company was inaccessible if you didn’t speak Vietnamese or they specifically had an English version. Nowadays a visitor can choose to get the entire site translated automatically (by the browser, for instance). Is it as good as the translation by an expert? Of course not. But it costs the company nothing at all and the visitor a negligible amount. And it works for a plethora of languages.
That’s another (invisible) way that the world needs less and less translators. I wrote this post in English but for all I know someone could be reading it in French or Bengali. No further input required from my side.
Maybe I’m not super up to date on AI stuff, but I worked as a translator for a year, and AI (they used ChatGPT and DeepL) still made a bunch of mistakes that you’ll immediately notice when you speak the language. It feels like their training input had a bunch of older, Google-translated articles in them that were just bad. Maybe an AI trained specifically for translation with curated learning material and a “teacher” who corrects mistakes can get closer to replacing human translators, but it’d still miss the cultural context of certain words and phrases that are in a translator’s passive vocabulary, at least in less widely spread languages.
That being said, it’s definitely harder to make a career out of translating because companies who don’t know any better just use AI instead. As long as they get their point across (and make money), they don’t care about the finer details.
Sure, a skilled human is still better at the job. But you don’t always need to capture every nuance. And AI does it at the fraction of the cost.
I see this with lots of German product descriptions on big store fronts like Amazon. They often seem entirely machine translated. It’s not great, but “good enough” and serviceable.
Machine translation can also increasingly shifts the process from the sender of the message to the recipient. It used to be that the web page of a Vietnamese company was inaccessible if you didn’t speak Vietnamese or they specifically had an English version. Nowadays a visitor can choose to get the entire site translated automatically (by the browser, for instance). Is it as good as the translation by an expert? Of course not. But it costs the company nothing at all and the visitor a negligible amount. And it works for a plethora of languages.
That’s another (invisible) way that the world needs less and less translators. I wrote this post in English but for all I know someone could be reading it in French or Bengali. No further input required from my side.
CAT-tools such as Trados killed the market. AI is just the natural conseguence.