RAE vs AI: the language battle nobody asked for

Published on May 29, 2026 | Translated from Spanish

The Royal Spanish Academy has expressed its concern about the impact of artificial intelligence on language. Its defense of clear language is valid, but it clashes with its institutional slowness and a historical elitism that has little to do with how people write on social networks or chats. The problem is not only technical, but social: AI replicates biases if it is not trained with real diversity.

Real Academia Española building interior, antique bookshelves clashing with holographic AI interfaces, a quill pen and open dictionary on a marble desk while a glowing neural network processes slang from a smartphone screen, linguistic data streams showing biased training sets, dust motes in dramatic light, cinematic photorealistic visualization, tension between old and new, ultra-detailed textures, architectural symmetry, warm amber tones contrasting with cool blue digital light

Algorithms that speak like your boss, not like your neighborhood 🤖

The real technical challenge is democratizing training data. Large language models often prioritize variants of Spanish from academic elites or dominant regions, ignoring idioms, slang, and forms of expression from marginalized communities. If linguistic diversity is not demanded in algorithms, AI will perpetuate inequalities: it will speak like a school textbook, not like people on the street. The solution lies in digital education and inclusive datasets.

The RAE correcting your WhatsApp since 1713 😅

While the RAE debates whether algoritmo has an accent mark, AI is already writing poems in Spanglish and responding to memes in inclusive language. The academy arrives late, as always, with the dictionary under its arm and a sour face. If they truly want to save the language, they should join the school group chat: there they will see how Spanish survives without their permission. Ironies of progress.