Can we bring together the devil, the king of Rome, the trolls, and Cao Cao?
Language is more than just words, we always knew that. It’s a fundamental element of the culture we feel part of, and it describes who we are. Think of all the idioms, proverbs, or just references to ways things work in the society or communities in which the language is spoken. Ever heard “speak of the devil (and he shall appear)”? That saying exists in a lot of languages but the “what” (you speak of) is different and might say something about who the group has feared. While German just like English speaks of the devil, Spanish speaks of the king of Rome. In Zulu, the King is in the air and in China, you speak of Cao Cao, a Han dynasty general. I learned all this by asking ChatGPT about expressions.
And while these common references keep a group speaking the same language together, it’s a barrier to the rest of the world. We can learn the words, even foreign grammar quite well, but those strange expressions that nobody can explain the logic of, we never really master in another language. We might mix up the devil and the king.
With ChatGPT and similar tools, we are now getting the opportunity to overcome those language barriers truly and fully. That is huge, and there might be existential consequences of this development that we will start to see over time but let’s focus for a minute on some very practical advantages. For instance:
More equal education: The fact that any text can be translated into a wide variety of languages in seconds opens up for much greater access to knowledge and sharing of it between groups of people. Training that is taken for granted in mature markets can be achieved in emerging markets too. And the other way around, the ability to learn about other cultures, about other parts of the world and about other topics than the main curriculums at schools will increase enormously.
Better quality of global research: This is something we as researchers always struggled with. How do we make sure the questions we write in English will be asked in the right way across all the languages we’ll use in global research? The questions we ask are very contextual so even skilled translators might struggle, and deep QA requires time we may not have. Chinese for instance, as we’ve learned the hard way, is very specific in each word’s meaning and translations can go very wrong. A well-trained AI translator could improve the quality and speed enormously.
Language preservation. Smaller languages constantly face the threat of extinction if those speaking find it better to move more of their conversations to other languages. But AI can help keep languages alive, at least that’s the objective of Open AI’s collaboration with the Icelandic state to preserve the Icelandic language as well as taking a step towards creating resources that could serve to promote the preservation of other low-resource languages.
This is all very positive, but if language is a key component and bearer of our identities, what does a world with equal access to any language do to our identity? This will be a very interesting field of research over the coming years. We may take the direction of moving further towards one joint culture to identify with across the globe, similar to what TikTok videos already provide young people around the world with a shared set of references, cultural moments, and experiences. But there is likely to be a counter-movement as well, going deeper into our roots, to stress even harder where we came from. And in parallel, we might focus on growing other features of identity, to replace languages as they become too generic. Things like hobbies or areas of interest, skills, the place we live in, or where we’d like to go. Time will tell.
