Congratulations, linguistic geniuses! 🎉 You’ve saved the English gentleman from the iron’s embrace! 🦾❌
Of course, the correct answer is option #2! ✅
This is a classic mistake of literalism.
🎮 How do we treat these hallucinations in applied linguistics?
🧠 Attention Mechanism — we teach the neural network to “pay attention” to neighboring words. If there is an idiom in a sentence, the model must understand that it needs to be processed as a single indivisible unit.
📚 Parallel corpora — we “feed” the algorithm millions of pages of text where a human has already provided the correct translation, so that the AI can learn from examples.
📐 Vector proximity — in semantic space, the expression “pull someone’s leg” should be closer to the word “joke” than to the word “physical exercise.”
Applied linguistics is about making machines more human.
Do you want to be the one who teaches AI to understand humor and metaphors?