Undoubtedly the hottest topic in the industry right now is the use of AI.
With so much hype surrounding the subject, it’s important to take a step back and rationally evaluate the variety of opportunities AI presents for processes within language services. Having served many years as a technology partner and a translation service provider, we are in no doubt that modern language technologies present many new opportunities; but if these technologies are to ensure added value in the long term, they need to be integrated into existing processes in a way that is structured and sustainable.

With us by your side, you will get to where you want to be. Climb aboard – we have a tradition of innovation!

We utilise the strengths of LLMs and NMTs, integrating them efficiently into the translation process. DeepL and ChatGPT are two well-known tools. But what is the difference between them?

• DeepL is based on neural machine translation (or NMT for short). NMT models are trained for translation by drawing directly from vast volumes of multilingual data.
• ChatGPT, by contrast, is a large language model (or LLM). These models are first trained to predict the next word using monolingual texts. In technical jargon, we call this “pretraining”. A second step is then required for LLMs to specialise in translation processes. This method requires fewer examples than the NMT model, since an LLM already has a certain basic grasp of how languages work and it “only” needs to learn a new language.