Generative AI

What it is

IBM defines generative AI as AI that can create original content, such as text, images, video, audio or code, in response to a prompt. Text models (LLMs), image generators, voice models and video models all fall under the term.

How it works

1. A model learns patterns from large training datasets.
2. Given a prompt, it generates new output that follows those patterns.
3. Products wrap the model with interfaces, safety filters and connections to your tools.

Benefits

• Speeds up first drafts of copy, images, code and reports.
• Makes analysis of unstructured text practical at scale.
• Lowers the cost of producing variations and translations.

Limits and risks

• Outputs can be wrong, biased or close to training material, so review before publishing.
• Copyright (U.S.). In January 2025 the U.S. Copyright Office concluded that AI outputs can be protected only where a human author has determined sufficient expressive elements. Prompts alone aren’t enough. Using AI to assist, or including AI material in a larger human work, doesn’t by itself bar protection. Rules differ in other countries, and the Office’s report on training data was still a pre-publication version when we checked.
• Disclosure questions apply to generated media (see content credentials).
• Costs scale with usage, so track spend.

This page explains concepts. It isn’t legal advice. Ask a lawyer before relying on ownership of AI-assisted work.