Generative AI
It is hard to fully explain how Generative AI models work but in principle, they use a library of content to act as the foundations for their learning model which are tagged with identifiable features, terms, and information that is then fed into a central model that once these terms are searched for combine elements that feature these identifiable search terms into a model that can be drawn or generated using techniques such as the diffusion model to generate new content. In image terms, generative AI models typically redraw elements based on what they think is correct based on terms entered using natural language models to humanize the search data into an understandable format.
The Lego box analogy:
Generative AI for example could be described as a box of Lego bricks of various sizes, colors, and shapes each with its unique ID and infinite combination potentials, we are in essence asking for it to generate the instructions for the new model to be built and it is collating all bricks that fit the criteria or as near as possible. Using the rules the model is trained to follow it then combines these bricks into the desired form and tries to produce as many variations as possible but its vision is somewhat limited as if it is trying to redraw the model looking through a fogged-up window, it knows the rough shapes and composition and tries to redraw what it sees and generate something unique. The clearer the instructions provided the better the model produced can be but ultimately it needs the building blocks to begin with and the time to learn how these fit together.
IT IS IMPORTANT TO NOTE THAT NOT EVERY GENERATIVE AI MODEL WORKS THE SAME AND GENERATIVE AI CAN INCLUDE 3D, AUDIO, VIDEO, AND TACTILE REPRODUCTIONS