Definition
Generative AI describes learned systems that produce or transform media. The output may be text, image, sound, video, code or structured data.
Scope
The concept includes foundation models, latent spaces, diffusion, transformers, prompts, embeddings, datasets, evaluation, provenance and human editing.
Applications
Electronic Artefacts can use generative AI as a research topic and production method when outputs remain connected to sources, constraints and interpretation.
Limitations
Generative AI can obscure training data, authorship, rights and confidence. Responsible practice documents process and avoids claiming certainty where the system provides probability.
References
See NIST AI RMF, provenance, generative systems and augmented intelligence.