Have we embraced the uncanny valley?

‘Prompt Jockey’ - created by me, using MidJourney with the prompts ‘prompt, jockey’ 2022.

I’ve been following and experimenting with the assorted prompt-to-image tools for some time. To say that this area has evolved quickly feels like an understatement. 

If you’re in this space you will know that there are now so many images that are being created with tools like Dall-E, Stable Diffusion, Midjourney and others and the return on prompt can be really interesting. There’s gallery upon gallery and of course these are being mixed with NFT technology to make these digital artworks available for purchase and ownership. It feels like a turning point in making art and as someone with a mix of artistic and AI skills, it feels like a perfect little storm as adverts, album covers, comic books and more are being produced, my sense is that this is at the start of its hype-cycle though, worth watching but due to go through more changes before it settles. 

Recently, while listening to a talk by an established traditional artist about his work using MidJourney, I was staring at one of his images - a picture of a woman. In a stereotypical sense, she looked very beautiful, clear skin, large eyes, a bit like a model or an actress. But something was a bit wrong - as it tends to be with these images. We can fish out a sense of what an image is meant to be and when those images are clearer, we can easily gloss over the imperfections of scale, shading and perspective. 

It got me thinking about the old uncanny valley - a place that I now feel is almost a mythological area in our minds. It’s usually related to images of people and considers our emotional response vs the image’s degree of accuracy in resemblance. When this is close but not perfect, the sense of the uncanny has been deemed unsettling. It’s usually applied to digital images of virtual actors and sometimes to humanoid robots. 

‘Uncanny Landscape’ - Image by me using MidJourney using prompts ‘Uncanny, landscape’ 2022.

As AI-generated images creep into popular culture, it feels like we are starting to embrace the uncanny. We’re getting to know images that are almost real and filling the gaps with our emotional responses and perceptual restoration (our minds and hearts completing the pictures). 

Technology often swings from hot to cold in its development and popularity and the development of one thing can be a real springboard for something else entirely. (think microchips and mobile computing). If we get to know the uncanny valley, maybe in areas of tech where this has been a sticking point, it will become a green light for faster development. As we accept avatars, will we equally accept avatars that are almost real? Virtual influencers like Lil Miquela point to this happening already. 

I am curious as to how we will see the world around us too. I’ve already seen a tree trunk that looked like it had been processed by Deep Dream. Maybe we will start to use stylistic vernacular relating to AI image creation prompts to describe our worlds IRL.