Generative AI, explained
IBM defines generative AI as AI that can create original content in response to a user’s prompt or request. Google’s machine learning glossary calls it an emerging field with no formal definition. It says most experts agree these models make content that is complex, coherent and original.
Much of the AI businesses used before was predictive: it sorted, scored or forecast things from past data. Generative AI makes something new, like a draft email or a product photo. A person should still check the output before it goes out, because it can sound sure and still be wrong.
What are examples of generative AI?
Chat tools that write text, image tools that make pictures from a description, and tools that make speech, music or video. Google’s glossary lists large language models and image, audio and video models as examples.
How do businesses use generative AI?
For first drafts of emails, posts and reports, for summaries of long documents, and for pictures and video made from material they already have. The safe pattern is the same each time: the tool drafts, a named person checks, and only then does it go out.
Questions
Is generative AI the same as an LLM?
No. A large language model is one kind of generative AI, the kind that writes text. Generative AI also covers models that make images, audio and video.
What is the difference between generative AI and an AI agent?
Generative AI makes content. An AI agent uses an AI model, often a generative one, to choose and take steps toward a goal, such as checking a calendar.
Can generative AI make mistakes?
Yes. It can write text that sounds right but is false, which Google’s glossary calls a hallucination. Check facts, numbers and names before you use the output.
Where we use this term
Sources
- What is generative AI?. IBM.
- Machine Learning Glossary. Google for Developers.
Definition checked against these sources and our own pages on .