Generative AI

Artificial intelligence that generates new content, such as text, images, audio, or code, by learning patterns from data.

Transformer: A neural network architecture that uses attention to process relationships among elements in a sequence. Transformers power many text generators by predicting tokens in context.
Machine learning: A field of computing that develops systems which learn patterns from data to perform tasks. Generative AI is one way machine-learning systems use learned patterns.
Large language model: A language model trained on large text collections to predict and generate sequences of words or tokens. This model class underlies many text and code generation tools.
Discriminative model: A model that predicts labels or outcomes from input data rather than generating new examples. It distinguishes prediction among known outcomes from content creation.
Copyright and artificial intelligence: Legal questions about copyright in AI training data and machine-generated works. Training and output raise disputes over permission, authorship, and protection.
Diffusion model: A generative model that learns to reverse a process of gradually adding noise to data. Many image generators create pictures by denoising random patterns step by step.
Neural network: A computational model of interconnected units that learns representations by adjusting numerical parameters. Neural networks are the main model family behind contemporary generative systems.
Text-to-image model: A model that generates images from written descriptions. It makes language prompts a direct interface for visual generation.
Search engine: A system that retrieves and ranks existing information in response to a query. Retrieval returns existing sources, whereas generation composes new outputs.
AI alignment: The study of making AI systems behave in accordance with human intentions and values. A generator can satisfy a prompt while producing harmful or unwanted content.

Linked from 10 pages