AI Glossary
Generative AI
GenAI, generative AI, generative artificial intelligence
Generative AI is a class of models that create new content — text, image, sound, video, or code — from patterns learned in data, rather than merely classifying or predicting values for existing data.
- It generates new content, not just labels or forecasts the way analytical models do.
- It spans text, image, sound, video, and code, often in a single multimodal model.
- Large language models and diffusion models are the two most common families of generative models.
Generative AI is a branch of artificial intelligence in which a model does more than analyze existing data — it produces new, plausible content. Having learned the statistical relationships in its training material, it can write an answer, draw an image, compose a sound, or generate code that was not present in the input data.
The key difference concerns the model's objective. Classic, analytical machine learning answers questions like "which category does this case belong to" or "what will the sales figure be" — it returns a label or a number. A generative model instead produces a new sample from the learned distribution, that is, the next word, pixel, or tone, step by step.
In practice, two families of models dominate enterprise use: large language models for text content and code, and diffusion models for images and graphics. Increasingly, both approaches are combined in a single multimodal model that takes in and produces content across several formats at once. From a deployment standpoint, the crucial point is that the output is probabilistic — each run can give a different result, which calls for quality control and clear limits on where it is used.
Related terms