What Is Generative AI?

Generative AI is the class of artificial intelligence that creates new content — text, images, code, audio, and video — by learning patterns from massive datasets and producing novel outputs on demand. ChatGPT's launch on November 30, 2022 made it a mass-market phenomenon, reaching an estimated 100 million users within two months, the fastest consumer application adoption recorded at the time.

What sits under the umbrella?

Generative AI spans several model families. Large language models (GPT, Claude, Gemini, Llama) generate text and code. Diffusion models generate images and video. Speech models synthesize voices. All share one architecture lineage — the transformer, introduced in the 2017 paper "Attention Is All You Need" — and one training pattern: learn from web-scale data, then generate outputs that continue or answer a prompt.

Where search and brand discovery fit

The commercially decisive application for marketers is generative search. Answer engines — ChatGPT Search, Perplexity, Google AI Overviews and AI Mode, Copilot — replace the ranked-links page with a single generated answer, often naming specific products and citing a handful of sources. Discovery moves from "which links rank" to "which brands the model names and which pages it cites." That shift created Generative Engine Optimization as a discipline, along with a measurement stack of prompt tracking, citation monitoring, and share-of-voice analytics.

The map, briefly

Why the definition matters in practice

"Generative" is the load-bearing word: every answer is produced fresh, sampled from a probability distribution, grounded in whatever was retrieved at that moment. Outputs vary run to run, engine to engine, and model version to model version. Brand presence in generative systems is therefore a distribution to be measured — repeatedly, across engines — not a rank to be looked up once. The glossary entries on answer volatility and prompt sampling cover that measurement problem in depth.