Term

Natural Language Processing (NLP)

NLP is the part of AI and computer science that deals with computers handling human language: reading, writing, classifying, translating, reasoning over it. The field moved through three eras: rule-based systems (grammars, lexicons), statistical methods (n-grams, hidden Markov models), and now transformer-based foundation models. Two narrower terms still show up in vendor decks: NLU (understanding only) and NLG (generation only). Modern LLMs do both.
Reviewed by 7wData
Mentioned Large Language Model (LLM)

Why it matters

Most things shipped under the label “AI” in 2026 are NLP deployments wearing different costumes. Email classification, sentiment analysis, summarisation, translation, chat, search, RAG retrieval, code completion: all NLP tasks. Foundation models collapsed the field’s task taxonomy. One model now handles dozens of jobs that used to need separate pipelines. That collapse is the real shift, not the chatbot interface. The older subfields have not disappeared: tokenisation, NER, parsing, and evaluation metrics still matter when you need structured output, want to debug a failure, or have to defend a result.

Where you’ll encounter it

Three concrete places. A vendor pitches a “vertical AI solution” and almost always it is an LLM finetune plus a prompt template doing a classic NLP task. A model-risk review asks you to separate “language understanding” from “language generation” because the failure modes differ (hallucination is a generation problem, misclassification an understanding one). A job description lists “NLP experience” and the meaning depends on the team’s age: in a 2018 team it meant statistical methods and Stanford CoreNLP, in a 2026 team it means transformer finetuning and prompting Claude.

Part of the 7wData AI Glossary. Tracking how concepts like this move in the expert conversation: daily signals at ins7ghts.com.