Generative AI fundamentals — AIF-C01

What it is

Generative AI is a type of artificial intelligence that can create new content and ideas — including conversations, stories, images, videos, and music. It works by learning patterns and relationships in training data so it can predict and produce novel outputs that are coherent and contextually meaningful.

The models underlying generative AI are often called foundation models: large models trained on a broad spectrum of text and image data that are capable of performing a wide variety of general tasks such as answering questions, writing essays, and captioning images.

Mental model

Think of generative AI as a completion engine at scale. For text, the model predicts the next word based on previous words and their context — repeatedly, until it produces a full, meaningful output. The same completion intuition extends to images and other modalities: the model has learned enough about the structure of the world that it can generate plausible new instances of it.

When to use it

The exam distinguishes generative AI from other AI approaches by what the system produces.

Capability	Generative AI	Other AI (e.g., classification, prediction)
Primary output	New content (text, image, code, audio, video)	A label, score, or decision from existing data
Example task	"Write a product description"	"Is this review positive or negative?"
Underlying mechanism	Learns patterns to generate novel sequences	Learns patterns to map inputs to predefined outputs
Typical model family	Foundation model / LLM	Supervised or rule-based model

Use generative AI when the goal is creating something new. Use a discriminative or predictive model when the goal is categorizing or scoring something that already exists.

Common misconception

A common trap is assuming that generative AI simply retrieves or recombines stored text — like a search engine returning pre-written passages. This is incorrect. Generative AI models learn patterns and relationships in training data and use those patterns to produce new content, not surface existing documents. The output is generated, not fetched.

A related misconception is that generative AI and "AI" are the same thing. Artificial intelligence is a broad field that includes smart assistants, image classifiers, robotic navigation, and much more. Generative AI is a specific type of AI defined by its ability to produce new content meaningfully and intelligently — it is a subset, not a synonym.

How it shows up on the exam

The cognitive target for this concept is recognition and contrast: candidates should be able to identify what generative AI is and distinguish it from adjacent AI types based on what the system produces.

Signal phrases to watch for in questions:

• "create new content," "generate," "produce novel output" — these point toward generative AI
• "classify," "predict," "detect," "recommend from existing items" — these point away from generative AI

Candidates sometimes confuse generative AI with traditional ML because both involve training on data. The grounding distinction from the official source is directional: generative AI generates new content and ideas; it does not merely analyze or label content that already exists.

Related concepts

• Embeddings — the numerical representations that capture semantic meaning, enabling generative models to process and relate concepts
• Foundation Models — the large, broadly trained models that serve as the basis for most generative AI applications
• Large Language Models — a prominent category of foundation model specialized for text generation and understanding