AI + art = ?
Some of you might have heard the news about Open AI’s Dall·E 2, a new AI system that can create realistic images and art from a description in natural language. Basically, you can type whatever you want into the command box and the AI system will generate a few images for you based on the language used (Dall·E 2 is an upgraded Dall·E, with greater resolution and more accurate images). For instance, you can start with the prompts on the site including “An astronaut playing basketball with cats in space as a children’s book illustration.”
You’d get the following. Pretty cool, right? And in the last few weeks, they released Outpainting, “a new feature which helps users extend their creativity by continuing an image beyond its original borders — adding visual elements in the same style, or taking a story in new directions.”
When playing around with it, I typed in “a piece of art that makes you feel good in a cartoon style with neon colors” (because why not?) and here are two of the images I got:
I enjoyed the clever results, and it was definitely fun to experiment with.
Access to the system had been waitlisted until yesterday, as Dall·E 2 wanted to regulate how many people have access and monitor possible repercussions (including misuse) from having the ability to create such images instantly. I don’t doubt that OpenAI’s intentions are good and the technology is awesome. In addition to generating brand new images, Dall·E 2 can also expand existing images creating new compositions or make realistic edits to add shadows or texture. They also had artists integrate this technology into their own practice too. If you’re curious about the more technical explanation behind how the system works and what the process of diffusion entails, I encourage you to check out the website.
All of the recent advancements have many people debating what the future of art will look like and what the role of AI is in it. Here at soal, we are building AI in a different way, using personalization as a tool to help make finding art you like easier. While the technology is undoubtedly impressive and increases people's ability to tap into their creativity, I have been seeing a lot of tweets and comments about how this type of AI-generated art is better than human production. Why pay more and wait longer for art created by artists when you can generate your dream vision in seconds? Some made similar arguments when the first photographs started appearing almost two hundred years ago. Many even thought it would be the end of painting as we knew it. But human emotion, personal experience, and that ‘human element’ mattered more to the picture. That’s why we at soal are so passionate about these personal and emotional elements when it comes to art. Dall·E 2, Stable Diffusion, Nightcafe, StarryAI—all of these are amazing tools. They can create and recreate, but can they truly innovate with art?