Aesthetic Artificial Intelligence

Luke Rexing, Contributor

Artificial intelligence has been sweeping the globe increasingly generation over generation. Some will never interact with this idea, for the fears that it will eventually surpass and dominate human existence. 

“I think that it is going to take away a lot of jobs and put people out of business.”

— Marko Mirkovic

Others are entranced by the idea of artificial intelligence, and push it to its limits to see what it can create for us. There are nearly infinite possibilities with AI, including self driving cars, playing chess, and being able to learn how to perform previously human-governed tasks. These are difficult ideas to process, as we don’t want a terminator situation on our hands, but there is something beautiful about artificial intelligence that has recently been released to the public. 

DALL-E 2 is a machine learning model developed by OpenAI to generate images based on natural language descriptions. In English, this program can essentially create any image so long as you can describe it. It is an artistic form of artificial intelligence, and is extremely fun and free to use. There is a waitlist, but it doesn’t take too long to get on the list. I signed up back in August and was able to get on the list by early September. I introduced it to my dad, where he has spent many of his credits creating images of french bulldogs (like my dog Winston). 

When you first sign up, you get 50 credits, or 50 text submissions for free during your first month. After this, it goes down to 15 credits per month. For being absolutely free, I don’t see a problem with this at all. It has been fun experimenting with what the AI thinks of relatively philosophical ideas. For example, this is the image that was created when I entered the text:

“The edge of human imagination”

This program is simply mind blowing. Their mission is to empower people to express themselves creatively, and I think that they have done nothing short of an outstanding job. I encourage other people to try it out to see what their imagination can come up with. Another thing that seems to be very well done by this program is its ability to generate images of animals, extremely detailed. The images created look like they have been taken by a camera, but have actually been generated based on machine learning. After seeing my dad try out many versions of french bulldogs, I decided to try one of my own.

This is the image that was rendered when I used the description:

“Close up of a proud looking black french bulldog wearing a top hat, studio lighting, backlit ears”

He looks exactly like Winston! In real life, we would never be able to get him to sit still enough for a picture like this. And he has a top hat! 

DALL-E 2 is super fun to use, and confronts many concerns some people may have about an image-generating artificial intelligence. OpenAI uses algorithms and screening techniques to filter out any description that is deemed violent or sexual. As it is open to the public for free access, these bases have been covered to ensure that nobody is inappropriately using this program. 

Although it may not get everything right all the time, for example, not being able to differentiate between things like “Latte art of a panda” and “A panda making latte art”, it is still extremely detailed. Users can even describe the type of style they want their image in, saying things like “in the style of Van Gogh” or “in the style of Picasso”. There are so many opportunities for something beautiful to be created, and it is interesting to see that it is coming from something that is not entirely human. I encourage you to sign up and try it out for yourself. Who knows, maybe you’ll be the next digital Banksy!