Image generation AIs like Midjourney or Stable Diffusion are attracting new users in millions. But why?
I have to admit that I'm probably addicted. I am addicted to throwing more crazy prompts for an AI to chew on so that it can provide me with more fantastical imagery. And by looking at the numbers, I’m not the only one. Millions of people are on the same hook, creating ever more fantastical images every day.
One of the most popular image generation AI services, Midjourney, has 3.8M users in their Discord server as of now. Instagram's hashtag #midjourney has 1.1 million images currently. And that’s just one of the many creative AI services that have popped up since late summer.
But what is it that makes image generation as a genre of AI so damn addictive to users?
It empowers your imagination.
One thing is that AI allows us to explore our imaginations in ways we could never before. We can come up with the most insane ideas, and the AI will create an image to bring them to life.
The imagined worlds fascinate us as humans. Maybe that's the main, if not the only, thing that sets us apart from the rest of the animals. Stories have been with humanity since the dawn of time. Stories that have taken us to unknown lands. Stories about heroes with superhuman abilities. Stories about villains and monsters that have terrified us and kept us awake on countless nights.
While most of us enjoy such imaginations, not all of us are capable of expressing our dreams as well as others. The further we go back in time, the more privileged the position of storytelling has been. What were ancient Egypt's priests, if not the world's greatest storytellers?
Up until now, however, this kind of visual storytelling was simply not accessible to laymen like me. But now, AI services like Midjourney, Dall-E, and Stable Diffusion have broken through that barrier. And AIs are improving really fast. They are starting to generate images that are photo-realistic, introducing textures and lighting that are nearly impossible to achieve in real life.
It's democratic.
Over the course of history, the power of storytelling has become more accessible to everyone. With the evolution of technology, more and more of us were able to create images, music, and texts. We can perhaps call this the democratization of the realm of creativity.
The creative AIs have just massively expanded "the Democratic Republic of Art". Millions of office rats, car mechanics, and housewives—they are all now almost as skilled artists at expressing their imaginations as the handful of people trending in Artstation.
It's visually captivating
Secondly, human brains prioritize visual input. The human eye can process 36,000 visual messages per hour. Visual input takes up 90% of the brain’s information processing bandwidth. It has been said that people process information in images 60,000 times faster than text.
The story that may take pages to tell can be told in one fantastical image. While the writer needs to take long passages to describe the atmospheric lighting, fog seeping between ancient trees, and sorrow in the face of the old man, it takes just seconds to convey in the case of a well-designed image.
This takes the whole "visual storytelling" thing to a new level. Not only are we free in our imaginations, but the output is somehow extremely realistic. And this kind of realism further amplifies the immersive effect of such imaginations.
It can be pleasingly realistic.
I believe there’s a certain overcoming of the "uncanny valley" effect going on. As AI-generated images are getting more and more realistic, we are beginning to like what we see.
The "uncanny valley" is a hypothesis in the field of robotics and 3D computer animation that suggests that the closer robots and other objects get to appearing like real human beings -- or at least like something we strongly empathize with - the more likely we are to have a strong, and mostly creepy, reaction to them if they fail to meet certain similitude levels. The recent advances have brought us to a territory where robot-generated images of humans do not look weird or repulsive to us.
Also, AI has made it easier to anthropomorphize objects and animals. In my own Instagram feed, the most popular images are a bunch of frogs in gym hoodies. One of the most popular Midjourney themes is fantastical anthropomorphized creatures: warrior lions, mystical cats, raccoon ninjas, etc.
Maybe this is going on because we as humans feel lonely in the universe, and it would be totally awesome to meet someone with the same faculties as we have but from a different species. Who knows? But one thing is sure: people like that, and nothing will stop them from generating another image of the Parrot Queen of the Jungle that somehow resembles Jessica Alba.
It connects.
The drive to generate more fantastical AI images may also be related to the feeling of loneliness from another angle. We want to be a part of something, and AI art communities on social media are springing up to meet that need.
We can find a lot of like-minded people on different social media platforms, who are just like us: amateur image creators. We follow each other, like each other's images, and tell visual stories to each other.
I won’t argue that it works to overcome loneliness. Research has demonstrated that broadcasting your images on Instagram may actually be counterproductive. But, hey, to hell with that research if I can get another 10 likes from total strangers to my albino frog in the Adidas hoodie.
***
In conclusion, image creation AIs are so damn addictive because they help us, the idealists by nature, bring our imaginations "to life," and they have democratized access to visual storytelling.
We love those visual stories, because we get so much more information out of them, compared to the texts.
The amazing level of realism the AI models have achieved has helped to overcome the "uncanny effect" and produce likeable images of real humans we can relate to.
And furthermore, they produce anthropomorphic frogs, lions, cats, dogs, and wolves that we can relate to too. And finally, we can share all those creations easily and hope that perhaps someone likes them.