Artificial intelligence and foreign languages: why learning languages matters more than ever
There is a misunderstanding that needs to be cleared up straight away. When we say ‘artificial intelligence’, we are using a term that creates a misleading parallel with human intelligence. The word ‘intellect’ derives from the Latin *inter-legere*: to read between the lines, to grasp the deep connections between things. AI, however powerful, does not read into anything: it recognises statistical patterns, produces probabilistic responses, and excels at repeatable, well-defined tasks. But it stumbles over irony, ambiguity and double negatives – over everything that makes human conversation something irreducibly complex. Intelligence, in the full sense of the term, is something else entirely: it is the ability to grasp the meaning of a context, to read the relationships between things, and to act judiciously in the face of uncertainty. As Father Paolo Benanti has observed, the Greeks already distinguished between metis – practical intelligence, the kind that finds solutions to problems – and nous, the intelligence capable of grasping the meaning of the whole. AI excels at the former. The latter remains human, multifaceted, capable of moving in multiple directions. So, is artificial intelligence really intelligence? The question is not an idle one, especially for those who work with languages.
It is commonly believed that translation is one of the activities most vulnerable to automation and therefore most easily replaceable by AI. This oversimplification confuses lexical transposition with linguistic comprehension. Translating means immersing oneself in the context of a conversation, deciphering cultural codes, reading between the lines of non-verbal communication, and adjusting one’s tone to suit the interlocutor. No probabilistic model, however sophisticated, can replace this form of relational intelligence. It can support it, but not replace it.
The real issue is not about replacement but about enhancement. At the Stanford Institute for Human-Centred Artificial Intelligence (HAI), a vision has taken hold that turns this perspective on its head: not Artificial Intelligence pitted against humans, but Augmented Intelligence at their service. The aim is not to automate human capabilities, but to enhance them. Not cost efficiency, but value creation. As is often reiterated in the contemporary technological debate: AI will not replace workers, but will automate certain tasks, freeing up time and resources for those with higher added value. Language professionals will not be replaced, but those language professionals who use AI will replace those who do not.
This is where the central importance of meta-skills comes to the fore: critical thinking, adaptability, the ability to learn how to learn, and emotional intelligence. These are the skills that underpin all the others, enabling us to navigate complexity and to integrate technical ‘hard skills’ – where AI is a formidable ally – with distinctly human ‘soft skills’: empathy, negotiation, leadership and cultural sensitivity. The challenge, then, is not to defend languages against AI. It is to train people capable of using both – machines and language – with the awareness that the responsibility for decisions, the building of relationships and the meaning of a conversation remain human. Because, as the Human-Centred approach reminds us, systems must serve people, not the other way round.
This is precisely why language teachers today play a more strategic role than one might think. They do not simply impart vocabulary and grammar: they cultivate the ability to navigate between different linguistic and cultural systems, to recognise otherness, and to negotiate meanings in unpredictable contexts. These are skills that a machine cannot develop on our behalf, because they require an embodied experience of language – cognitive, emotional and relational.
