What is it, what does the future hold?

The world of AI text to art generators has grown by leaps and bounds over just the last few months. Images generated by AI have won art competitions against humans, been granted copyright, been used to illustrate stories for major publications and likely worked their way into your social feeds, whether you know it or not.  Here’s what you need to know:

What is it?

AI text to art generators are algorithms programmed to generate art from just a short text description of the image. For example, I typed in “pick up truck, wheat field, daytime sun, photorealistic” and the computer spit out a photo approximation of a sun drenched wheat field with a truck parked in it.

The AI has been trained on millions of images scraped off the internet in order to know what things are and how they should look. This means the algorithm also understands the styles of human artists. For example, I typed in “Aliens at a kegger in the style of Van Gogh Starry Night, oil painting” and the computer will dutifully create an image of little green men standing around a keg drinking from red party cups, in Van Gogh’s signature impressionist painting style.

The current most popular text to AI art generators are Dall E 2, Midjourney and Stable Diffusion. Although they all have different interface designs and general vibes, they are all extremely powerful even though the technology is still in its infancy.    

The good:

AI text to art generators have gone from very basic to uncanny in ability in a very short period of time. They are likely to democratize the creation of many kinds of art in the near future with shocking rapidity. Millions of AI images are being generated from text prompts. Anyone is able to log in, write a short description of whatever is in their imagination and have it instantly appear. With some iteration and refinement, it will be easy to create anything you can imagine. The applications are almost unlimited. Currently, most of these image generators cost a few cents per image generated, but Stable Diffusion is open source and free to download, with the caveat that it takes some technical know-how to install. 

The bad:

AI may very well put some artists who have spent years honing their craft out of business. There are also legal questions about ownership and copyright, since the AI is creating the actual art as described by a human. In at least one instance, a copyright has been granted for a graphic novel in which all the images were created by AI, but written by a human. On the flipside, living artists whose styles are being mimicked by AI are raising legal questions about AI being “trained” on their bodies of copyrighted work. Stock image services and art websites have also been banning AI artwork in an effort to keep the sheer volume of AI images from overwhelming actual human created artwork and photographs. 

The future:

For better or worse, the era of text to AI art has arrived and is already eye-poppingly impressive, even in its early days. Innovation in the field is continuing to happen at a breakneck pace. Meta, the parent company of Facebook and Instagram announced a text to video AI just this week. Currently, it’s estimated that 1.5 million users are generating 2 million images a day. It seems only a matter of time before much of the content we consume online is generated by a computer from a text description.