Behind the Scenes of AI in Art

Abstract

The creative industry has been in turmoil since artificial intelligence (AI) has begun to generate its “own art.” But AI was not always this intelligent, there are many past models that are milestones in AI’s journey to image generation based on text prompts. Programs like OpenAI’s DALL-E* are able to generate an image based on a user’s text prompt, using a Contrastive Language-Image Pre-Training (CLIP) model, a diffusion model, and a prior model. Other programs created by smaller organizations or individuals, use some sort of image generator such as a generative adversarial network (GAN) plus CLIP to generate images based on text prompts. These AI programs have led to discussions regarding the legitimacy, or originality of these images, and the data images used to train the programs.