All you need to do is a bit of research... as gill said...

How does an AI generate images?......simply put 

To generate images, the machine uses two neural networks. The first neural network is used to create the image based on the text input by the user. The second neural network analyzes the generated image with reference images. By comparing the photos, it creates a score to determine the accuracy of the generated image.

As to why, lets say someone ask the ai to render megan fox...it will never ever look like her, it will be a close approximation, if it was actually using images...why not just grab one...then it would be...

A never ending debt...