I often find that the hardest part of writing a social media post is selecting a good accompanying picture. My go-to source for beautiful (and for the most part free) photos is Unsplash but sometimes it's hard to find the perfect image and, believe me, I can spend waaaay too much time looking! So recently, when trying to find a picture for my business birthday Linkedin post, I tried a different tactic, one that was inspired by my children who once whiled away a long train journey creating the most bonkers images they could: I turned to AI and tried out NightCafe.
Is it any good?
NightCafe provides multiple methods of AI art generation and is very simple to use. I used the Stable Diffusion text-to-image algorithm which allows you to create images from a text prompt. I played around with some of the different styles on offer (such as artistic portrait, photo, and pop art). It took me quite a few attempts to generate an image I liked but this was partly due to the lack of detail in my initial prompts. In some of the resulting images, there were “people”- utterly ghastly humanoid creatures that I very quickly deleted. In the end, I managed to create a basic image that worked for my post but it wasn't anything special.
Subsequent attempts at generating images produced more results that didn’t quite hit the mark. For example, I tried to create one of a frog on a plate to update my previous Don't eat that frog blog post but some of the frogs were very weird with elongated bodies or the wrong number of toes. Clearly, AI-generated art has its limitations but it's likely these kinks will be ironed out in the future. Also, judging by the images that some creators manage to produce, there’s skill involved in selecting the right prompts and modifiers to use (so much so that there are websites like Promptbase which actually sell prompts!).
Is it art?
I certainly thought of the results my prompts produced merely as “generated images” rather than “works of art” but I have to admit that some of the images others have produced are sublime. One such picture entitled “Théâtre D’opéra Spatial”, created using Midjourney, even won an art prize in America and this led to furious backlash. Many say that AI-generated images cannot be considered “real” art because AI lacks intent - it is not trying to express itself, or convey its culture, and has no emotion or lived experience to draw on or communicate. It can be argued, however, that there is a human behind the process who provides the creative spark - someone who decides what prompts to input, what image to use, and who is trying to express an idea or convey meaning. As with photography, which was once regarded as the death of visual art, supporters contend that AI art will eventually be recognised as a legitimate medium through which people can express their creativity. NightCafe, certainly holds this position (albeit with a vested interest). They say that they are “on a mission to democratise art creation” allowing “anyone - regardless of skill level - to experience the satisfaction, the therapy, the rush of creating incredible, unique art.”
Artists like Steve Coulson, featured in this article, would agree - using AI allowed Steve to fulfil his dream of publishing a comic book, despite not being able to draw!
Even if it is art, is it ethical?
Creating images with AI is undoubtedly cheaper (and often free) and as such could affect the livelihoods of many creatives. Having said that there are those who have used AI images as inspiration for their art or writing and, as the technology advances, AI art generators may present a cost-effective and accessible way for small companies to create marketing content and allow them to easily mock up ideas.
A related problem is that AI art generators “learn" by scraping images from the internet, even those that are copyrighted, essentially stealing the existing works of living people, without crediting or compensating them. It could be argued that this is simply what humans do and all art is derivative - humans look at other works of art which inspire them to create their own. But many such programs allow users to create images in the style of a particular artist. Some artists, like Greg Rutkowski, featured in this article, have gained recognition this way but at what cost? His name has become one of the most commonly used AI art prompts but Greg isn't happy about it as he fears that his own work will be totally eclipsed by the AI creations tagged with his name but not of his making. There are also data and privacy concerns as it seems that for most of the AI art generators, artists don’t have the choice to opt in or out of having their work used this way.
Should we use it?
AI art is gaining in popularity - even Canva now has a text-to-image AI image generator and many people are enthusiastically embracing the technology but should we use it? Perhaps somewhat optimistically, I do not think that AI art will ever entirely replace art created by humans, just as photography has never entirely replaced painting. There are clearly some legitimate concerns though. When I first started experimenting with NightCafe, I viewed it simply as an image-generating tool, and a bit of fun. It was only when I did a bit more research that I became aware of the ethical concerns people have. Personally, I feel that developers have a moral duty to ensure that the images being used as datasets are ethically sourced with artists given control over how their intellectual property is used. I also think the industry should somehow be regulated. Hopefully, we will see this happen in the future.
Do you use AI image generators? Do you think the results qualify as art? And do you think AI art is ethical?
First published 9/11/22. Edited 26/2/24.
Cover photo by Tara Winstead: https://www.pexels.com/photo/robot-pointing-on-a-wall-8386440/