Creating text, images, and more
Generative AI makes content in response to a request. It might draft an email, create an image from a description, or write code. Other AI systems might predict whether a payment is fraudulent. One product can do both.
A large language model can generate text. Other models can create images, speech, music, or video. A multimodal system can work with more than one type of content.
The app shapes the result
The model is one part of an AI product. The app gives it the user’s request and instructions, may look up documents or use other software, and decides how to show or use the result. Some actions may need a person’s approval.
Generated content can sound convincing and still be wrong. It may include made-up facts or reflect gaps and biases in its training examples. Teams should test outputs for their intended use and decide when a person needs to review them.