Anyone who has a bilingual career or speaks two languages at home - and there are many - knows that professional translation is no easy task. It takes time and, above all, skill. In addition to the widely available translation tools, artificial intelligence can also be used to help achieve the perfect result. With Lengoo, Christopher Kränzler and his co-founders have developed such a tool: The software uses machine learning to translate subject-specific texts.
We spoke to CEO Christopher Kränzler about what makes Berlin attractive, the advantages of using AI in translations and why we should definitely maintain our enthusiasm for learning languages.
Hello Mr Kränzler. You were born in Bavaria and after having worked in Karlsruhe, Frankfurt and New York, you ended up in our beautiful Berlin. What do you like so much about Berlin?
Berlin is an ideal breeding ground for innovative start-ups and attracts people from all over the world. For us as a fast-growing tech company, this is a major location advantage. Qualified specialists are rare and we currently need a large number of data scientists and machine learning engineers to further develop our translation technology.
In 2014, you co-founded Lengoo in Karlsruhe. Your company produces software that can deliver subject-specific translation through sophisticated machine learning. How exactly do you train your AI?
One of the unique features of our machine training concept is the high degree of specialisation. We train the language models for neural machine translation with the previous translation data of our customers. In this way, we individualise the AI systems and adapt them to the linguistic subtleties, the preferred terminology, the style and the tonality as well as the intended use. Thanks to this specialisation, the machine output achieves such a high level of quality that allows professional translators to work three to six times faster with it. This enables us to cut the costs of specialist translation by half compared to traditional translation providers. On average, this saves the customer a mid six-figure sum in their annual budget. This effect continues to gain momentum, because the machine continues to learn, with the final corrections made by humans.
How does the use of AI make your software stand out from conventional translation services?
Firstly, the use of AI makes us faster and secondly, we ensure that the customer's language style remains consistent in every translation. Traditional providers simply cannot offer that. There are usually several translators working for one customer and because people are different, the results also differ in the end. For each word in the source language, there are different translation options to choose from. Our technology learns the individual corporate language style and thus ensures consistency in the translated texts.
At what point is human support still required? Does the AI translate everything on its own or do humans and AI work hand in hand?
Neural machine translation, as we use it, provides translations on an almost human level. To ensure that the machine output is 100 per cent correct, we need people in the process. Our professional translators are quality inspectors and, at the same time, machine teachers. Only when they interact with our AI system can the neural networks continue to learn and improve. I would say that man and machine complement each other perfectly in our translation process. As already mentioned, the technology can ensure a high level of consistency. Humans can recognise subtle linguistic nuances and ensure that they are correctly reflected in the target language. For this you will always need language experts who have a very good command of the source and target language and are familiar with the respective subject area.
Do we still have to learn languages in the future when an AI can always translate everything directly for us?
Of course, technology will make it easier for us to communicate in other languages in the future. However, machine translation will not replace language understanding. The pocket calculator doesn’t relieve us of logical and abstract thinking. I therefore believe that as communication becomes more technologically advanced, it will become even more important to raise awareness of the complexity of languages and to acquire intercultural skills
You recently raised $ 20 million in a financing round, congratulations! What are your plans for the near future?
Thank you! We want to use the additional capital to build a global presence for our international customers. We also aim to drive the development of our own NMT framework to make professional translations even faster and more scalable.
Berlin is the leader in AI start-ups and machine learning development in Germany. What else do you wish for in the future? Where can we take a slice of the US, China or other markets?
In my opinion, it is a matter of these developments in machine learning finding their way into application. If we manage, for example, to also introduce AI into German medium-sized companies, this would achieve a great deal. From a global perspective, we, as Europeans, now have the opportunity to position ourselves as world leaders. We can only achieve this with a positive attitude towards technology and in that respect we could adopt a bit of the enthusiasm seen in the US and China. I hope that we in Germany and Europe will increasingly recognise the advantages of artificial intelligence and thus allay some of the fears about the technology.