Artificial dreams
Do algorithms dream of electric sheep?
What is creativity? Can a machine, an algorithm, be creative?
A dictionary definition would have creativity as “the use of imagination or original ideas to create something.” Can a machine use imagination? Can an algorithm create original ideas? Good luck getting out of that rabbit hole.
Why not drop the focus on the process and, instead, pay attention to the outcome? The “imagination” and “original ideas” suggest a human-centric view of creativity. Does this view still hold in the world where we’re increasingly surprised by what technology can create?
I decided to play with “machine creativity” myself. Inspired by Philip K. Dick, I asked: do algorithms dream of electric sheep? (I didn’t have any androids around, so algorithms had to suffice.) What resulted is a gallery of 12 graphical works, all dreamed up by an algorithm.
I used the recently released GLIDE AI model to generate all the images. The only human work involved was me asking the algorithm to produce an image and then framing and hanging the pictures in the virtual gallery above. For instance, the algorithm created the image on the far right after I had asked the algorithm to generate “two electric sheep grazing on top of a motherboard”.
The technology used to create these images didn’t exist a couple of years ago.
I wrote a Twitter thread that contains all images and text prompts. It has a bonus image, just for Twitter users, and misses one from the gallery above—the one on the far left, “a dog in the style of keith haring”. I needed a herding dog for my sheep. Also, I like Keith Haring.
I encourage you to click on the top image or go to the Twitter thread and have a closer look at the works. Every image took at least a few minutes to be dreamed up, and each one surprised me with its creativity. I found myself eagerly typing the text prompts and then curiously waiting to see what unfolds.
If you can describe it, it will exist.
And I′ll let go of all that I know
Dead dreams and submarines adrift in the cold
And I'll find home at the end of the road
Waiting to see what unfolds
There are millions of ways to achieve a creative outcome. And so, thankfully, creativity research drops the process focus and pays attention to the result: “creativity involves the production of novel, useful products”. Are the images in my gallery novel? Without a doubt! Are they useful? Judge for yourself; I know: they helped me write this newsletter.
A few weeks ago, Twitter lit up for a while, sparked by a beautiful music video. The viewers described it as “lucid dreaming while wide awake”, “healing”, and “haunting”.
Almost no one knew the singer and songwriter, Zia. It was her debut single. And not many knew the video director, Aza Raskin, either. But the most surprising was the visual artist behind the video: an algorithm. Aza worked with the algorithm in a similar way I worked on the images in the gallery: by telling it what it should create (“sad and dark night sky with stars and the moon”, “clocks as faces under water, salvador dali”), and then combining the results into one piece.
The whole process took Aza less than 48 hours. Take two minutes to watch and enjoy it.
Aza himself was amazed by the outcome. He wrote: “the song is dark and tinged with hope, just like this tech”.
He concluded by saying: “the rule will be: if you can describe it, it will exist.”
Welcome to the new era of creativity, where “I am not creative” is not an excuse anymore. A business model? A new marketing campaign? A new pair of sneakers? If you can describe it, it will exist.