Finding a common language - how language AIs bring people together

In an increasingly globalized and digitalized world, communication plays a crucial role in connecting people, cultures and ideas. However, despite technological advances, language barriers often remain an obstacle to exchange. This is where language technologies come in: With the help of artificial intelligence (AI), it is possible to translate texts, overcome language barriers and even support complex interactions in real time. These technologies are developing at a rapid pace and are not only changing our everyday lives, but also the way we communicate.

Prof. Dr. Jan Niehues, head of the “AI for Language Technologies” research group at the Karlsruhe Institute of Technology (KIT), is working intensively on the opportunities and challenges arising from the use of language AI. In this interview, he explains how modern methods such as deep learning are helping to push the boundaries of machine language processing and provides insights into where these technologies are already being used today. He also takes a critical look at the ethical aspects and potential risks associated with the development of AI systems. This interview offers an exciting insight into the transformative power of language AI – and shows how it can help people to understand each other better.

Machine Translation Market Size and Share Analysis – Growth Trends and Forecasts (2024-2029). Mordor Intelligence. Retrieved November 20, 2024: Source

karslruhe.digital (k.d): Prof. Dr. Niehues, what exactly is meant by language technologies in the context of AI and which fundamental technologies are used?

Jan Niehues (JN): Language technologies are methods that deal with the processing of natural language, i.e. languages that we speak, such as German, and not artificial languages, such as programming languages. Although it is relatively easy for humans to understand languages and resolve their ambiguities, this has long been a major challenge for computers. Although computers have been able to play chess better than any human for almost 30 years, they have long had great difficulty in understanding, for example, when a “sign” means the omen and when it means the sign in mathematics. To solve this problem, we are currently using procedures based on deep learning methods.

k.d: Language technologies such as large language models (LLMs) are increasingly permeating our everyday lives. Where do we encounter these language technologies in everyday life, perhaps without even realizing it?

JN: So-called “large language models” are being used more and more frequently as an aid in the creation of texts. This means that a significant proportion of the texts we read today were probably written with the support of language technology.

k.d: Speaking of texts – translators in particular resist the use of AI, arguing that translating literature, for example, requires a lot of knowledge about cultural context, humor and artistic input. What are the advantages of artificial intelligence?

JN: On the one hand, artificial intelligence has the advantage that it is cheaper and is available more quickly. It therefore enables the translation of texts that would not be translated at all without artificial intelligence. Or an example from the university: lectures cannot be translated live by a translator in order to provide easier access for students who do not yet speak German well enough. This is different with AI. Using the lecture translator, we can translate lectures at the university directly into another language.

k.d: Let’s stick with your example of a language barrier being resolved by AI. Where do you see further opportunities for language technologies for society?

JN: They can help us in many situations. We are suddenly able to read texts in foreign languages or talk to people who speak other languages. So they can already help us to become more accessible. They can help us get more information or cope with the flood of information. Like many technologies, they can of course also have a negative effect, e.g. in the creation of fake news. I therefore think it is very important that we learn more about the methods, opportunities and risks of these technologies.

k.d: Do you also see limits to the possibilities of voice AI?

JN: As I said, it can help us to communicate with people in other languages, it can help us to formulate letters, reports or articles faster and better. It will therefore also have a major influence on how we communicate in the future. However, this can also lead to documents that are very easy to read but whose content is not correct. Critically scrutinizing texts is therefore even more important and perhaps even more difficult.

k.d: Do you have any ethical considerations regarding the development of AI?

JN: As with most new technologies, there are of course also risks with AI that it could be used in a questionably malicious way, for example to generate fake news. Therefore, I think it is very important to be informed about these technologies and to be aware of their opportunities and risks in order to assess them correctly and to be able to better recognize where it is fake news. In addition, there is research that AI systems can reinforce prejudices. It is important to include this as a research question and develop methods to reduce these negative effects, as is currently the case in the research community.

k.d.: What has surprised or inspired you personally the most during your research work so far?

JN: I am always surprised at how quickly the field develops and how quickly we get better results. I also find it very interesting to see how people use these systems and improve their everyday lives with these tools.

Thank you very much for the interview.