Natural Language Processing: How AI Reads and Writes
The big idea: Computers do not understand words the way people do — they only work with numbers. Natural Language Processing (NLP) is the branch of AI that turns human language into numbers a machine can compute with, lets a model find patterns in those numbers, and then turns the model's numeric answer back into useful language: a translation, a summary, a chatbot reply, or a "this review is positive" label.
Every time you use Google Translate on a Kyrgyz menu, dictate a message instead of typing it, or ask a chatbot for homework help, you are using NLP. In this lesson you will see why human language is surprisingly hard for machines, and the clever trick — turning words into vectors — that finally made language AI work well.
By the end of this lesson you will be able to:
- Explain why language is hard for computers: ambiguity, context, idioms, sarcasm, and the sheer number of languages.
- Describe how NLP turns words into numbers (tokens and word vectors) and why similar words end up close together.
- Name the core NLP tasks: translation, sentiment analysis, summarization, question answering, chatbots, and speech-to-text.
- Explain why neural machine translation was such a big jump in quality over older methods.
- Trace the full NLP pipeline: text tokens numbers model output.
1. Why Is Language So Hard for Computers?
For a computer, arithmetic is easy and language is hard — the exact opposite of how it feels for humans. A calculator from the 1970s could multiply large numbers instantly, but until recently no machine could reliably tell whether "Great, another Monday..." is a happy sentence. Here are the five big obstacles: