1. Distinguish between forms of learning

The training process is the first and most important phase in the development of generative AI. It is during this phase that the AI learns to create new content from the data it is provided with. Most generative AIs are based on Deep Learning neural networks, which require huge amounts of data to learn how to produce coherent textual or multimedia content.

The training of generative AI thus relies on a data set that is representative of the target application domain. For example, to train a language model designed to generate text (such as GPT), a large and varied corpus of text must be provided. If the objective is to generate images, the data set will be made up of millions of images.

There are two main learning methods used in generative AI training: supervised learning and unsupervised learning.

In supervised...