AI that outputs new content (text, images, audio, video, code) rather than an estimate about something that exists. Its distinctive risks include confabulation, information integrity, intellectual property and abusive synthetic content 1; its evidence is groundedness, refusal and red-team evals and content marking.
- Developed in
- ch. 11, Predictive versus generative
- Chapters
- ch. 11, AI Defined
- Contrast with
- Predictive AI
- Source
- 1 numbered reference, listed below
Where it is used
7 chapters of the Body of Knowledge use the term. Each link opens the first section that does.
- 08 · Regulatory Map NIST AI RMF 16 mentions
- 10 · Reading List Regulation and standards 6 mentions
- 11 · AI Defined Four definitions, compared 3 mentions
- 13 · Risk Management Identifying risk: sources, factors and stakeholders 1 mention
- 14 · Development Data for training and testing 1 mention
- 21 · AI Laws Worldwide South Korea: the AI Basic Act 10 mentions
- 22 · Principles & Standards G7 Hiroshima Process 5 mentions
Related terms
Sources
- [1] Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1 (twelve risks unique to or exacerbated by generative AI; confabulation defined). NIST. 2024-07. https://doi.org/10.6028/NIST.AI.600-1 (verified: primary)
Definitions of legal terms paraphrase the cited text, which governs. Dated statements are as of .