AI art and AI-generated art are two distinct approaches to creating art with the help of artificial intelligence, but often confused as one. In this article, we explore the differences between the two, and how they are changing the traditional art world. We look into how AI art uses machine learning algorithms to enhance human creativity, and AI-generated art uses algorithms to create art without any human input. We also discuss the impact of AI on the art world, and how it is opening new doors for artists and digital creatives. This comprehensive guide is ideal for artists, designers, marketers, and anyone interested in exploring the intersection of AI and art.
Exploring the Differences Between Generative Art and AI Art Generators
Artists have been experimenting with artificial intelligence for years, but the practice has gained new levels of awareness with the release of increasingly powerful text-to-image generators like Stable Diffusion, Midjourney, and Open AI’s DALL-E. Similarly, the genre of generative art has gained a cult-like following over the past year, especially among NFT artists and collectors. But what’s the difference? Does the category of generative art also include art made from super-charged AI art generators, too?
From an outsider’s point of view, it’s easy to assume that all computer-generated artwork falls under the same umbrella. Both types of art use code, and the images generated by both processes are the result of algorithms. But despite these similarities, there are some important differences in how they work – and how humans contribute to them.
Generative art refers to artworks built in collaboration with code, usually written (or customized) by the artist. “Generative art is like a set of rules that you make with code, and then you give it different inputs,” explains Mieke Marple, cofounder of NFTuesday LA and creator of the Medusa Collection, a 2,500-piece generative PFP NFT collection. She calls generative art a kind of “random chance generator” in which the artist establishes options and sets the rules. “The algorithm randomly generates an outcome based on the limits and parameters that [the artist] sets up,” she explained. Erick Calderon’s influential Chromie Squiggles project arguably solidified generative art as a robust sector of the NFT space with its launch on Art Blocks. Since its November 2020 launch, Art Blocks has established itself as the preeminent platform for generative art.
On the other hand, AI text-to-image generators pull from a defined data set of images, typically gathered by crawling the internet. The AI’s algorithm is designed to look for patterns and then attempt to create outcomes based on which patterns are most common among the data set. Typically, according to Versteeg and Marple, the outcomes tend to be an amalgamation of the images, text, and data included in the data set, as though the AI is attempting to determine which result is most likely desired.
While both AI art generators and generative artwork rely on the execution of code to produce an image, the instructions embedded within each type of code often dictate two completely different outcomes.
The artistic process differs significantly between these two methods. Generative art is about exploring the possibilities present in a set of rules and code written by the artist, leaving room for chance and experimentation. AI text-to-image generators, on the other hand, are about training an algorithm to produce specific outcomes based on a given set of data. The artist’s role is to guide the AI towards a specific result and then refine it further.
While both methods have unique advantages, concerns also loom large over AI-generated art. First, the origin of the data used to train the system is not clear. Second, cultural biases can influence the outcome. We’ve already begun to see examples of cultural bias emerge through AI art generators, revealing their potential limitations and highlighting the ongoing need for artists to produce their unique work.