Natural language processing — AIF-C01

What it is

Natural language processing (NLP) is the technology that allows computers to interpret, manipulate, and comprehend human language. It enables organizations to analyze large volumes of voice and text data from communication channels and to gain actionable business insights from that language data.

Mental model

Think of NLP as a pipeline that bridges raw human language and machine understanding. Text or speech enters one end; structured meaning — intent, sentiment, named entities, translated text — comes out the other. The pipeline depends on machine learning models trained on large language datasets, and modern implementations use deep learning (specifically neural networks with transformer architectures) to recognize complex patterns across an entire sequence at once rather than word by word.

When to use it

The exam often asks candidates to match a business problem to the right AI/ML category. Use this table to keep NLP distinct from adjacent areas.

Situation	Correct category	Why
Classify customer-feedback text as positive or negative	NLP — sentiment analysis	The input is natural human language
Detect objects in a product photo	Computer vision	The input is an image, not language
Translate a support ticket from Spanish to English	NLP — machine translation	Cross-lingual understanding of text
Predict next month's sales from historical numbers	Traditional ML / forecasting	Tabular numerical data, no language involved
Transcribe a recorded support call	NLP — speech recognition	Converting spoken language to text
Generate a draft email reply	Generative AI (NLP-based)	Language generation is an NLP capability

Common misconception

Candidates often assume NLP is synonymous with deep learning or with generative AI.

NLP is a task domain — the goal of making computers work with human language. It can be accomplished with rule-based methods, classical machine learning, or deep learning depending on the complexity of the task. Deep learning and transformer models are one implementation path, not the definition of NLP itself. Generative AI represents a significant advance in NLP capability (moving from processing language to generating it), but NLP includes many non-generative tasks such as named-entity recognition, sentiment analysis, and machine translation that predate and exist independently of generative models.

A related trap: NLP is sometimes confused with all of AI. NLP is a specific subfield focused on language; it sits within the broader AI landscape alongside computer vision, robotics, and other domains.

How it shows up on the exam

The exam tests whether candidates can identify NLP as the appropriate category when a scenario involves understanding, classifying, or generating human language. Signal phrases to watch for include: sentiment, intent, named-entity recognition, machine translation, speech recognition, chatbot, document processing, and language understanding.

Candidates are commonly asked to distinguish NLP scenarios from computer-vision or numerical-ML scenarios. When a scenario involves text or speech as the primary input or output, NLP is likely the relevant category. When the scenario involves images or structured numerical data, a different AI/ML category applies.

A common misconception the exam may probe: NLP is not limited to English or to written text — it applies to any natural human language in voice or written form.

Related concepts

• Artificial Intelligence Overview
• Deep Learning
• Machine Learning Basics