What is generative AI?
Generative AI is a kind of artificial intelligence that creates new content, such as text, images, audio, video or computer code. It learns patterns from the data it was trained on and produces new material that resembles that data.
Also known as: GenAI, generative artificial intelligence
Researched and fact-checked by AI, with no human review. 5 sources listed below. How we verify
Last updated
What counts as generative AI?
Generative AI is a broad label for AI systems whose main job is to produce content. A 2023 explainer from Georgetown University's Center for Security and Emerging Technology (CSET) makes the point by contrast. Other AI systems classify data, group it or choose actions. CSET's examples of generative AI include image generators, large language models, code generation tools and audio generation tools.
U.S. agencies use a more formal definition. The National Institute of Standards and Technology (NIST) quotes one in its July 2024 generative AI profile. It describes models that emulate the structure and characteristics of input data to generate derived synthetic content. That content can include images, videos, audio and text. The FDA uses the same wording in an August 2026 discussion paper on medical devices.
The term overlaps with others. CSET says a large language model counts as generative AI when it is used to generate content, but not when it is used in other ways. It adds that no bright lines separate these terms.
What are the main risks?
The NIST profile lists 12 risks that generative AI creates or makes worse. One is confabulation, which it describes as confidently stated but false content. People often call this hallucination. Others include harmful bias, data privacy and information security.
How is it used in health care?
The World Health Organization issued guidance in January 2024 on large multi-modal models, a type of generative AI. It listed uses such as answering patients' written questions, documenting patient visits and supporting medical and nursing education. It also noted documented risks of false, inaccurate, biased or incomplete statements.
Regulators are now working out how to oversee medical devices built on the technology. The FDA's paper says such devices may accept open-ended inputs and give different outputs to similar inputs. The paper is for discussion only and is not guidance. The FDA set an October 19, 2026 deadline for feedback. The AI in Medicine Regulation Tracker follows these steps.
Sources
- What Are Generative AI, Large Language Models, and Foundation Models?, Center for Security and Emerging Technology (CSET), Georgetown University
- Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, National Institute of Standards and Technology (NIST AI 600-1)
- Considerations for the Regulation of Generative AI-Enabled Medical Devices: Discussion Paper and Request for Feedback (PDF), U.S. Food and Drug Administration
- FDA Seeks Public Feedback to Inform Regulatory Approach for Generative AI-Enabled Medical Devices, U.S. Food and Drug Administration
- WHO releases AI ethics and governance guidance for large multi-modal models, World Health Organization
