If you are into art, much like me, and have looked at Twitter over the last few months (in tandem with the majority of us), you may have witnessed what AI can create on the artistic front. ‘AI art’ is a huge umbrella term, in my opinion, as the emerging technologies that are gaining traction are non-identical. To elaborate, AI art is both creatively fascinating and vastly intelligent, but also inconsiderately repulsive depending on its format and use. Let’s dive into that!

The Creative Side of AI Art: MidJourney

Search #MidJourney on Twitter and expose yourself to the plethora of gorgeous and thought-provoking art. I say that because AI art has its ambiguity! If you have jumped on the opportunity to join Midjourney’s public discord server and tried their generous free beta for a limited trial (of around 25 queries), you can advocate that AI created solely from a bot has its limitations. All good artists still use reference images – it’s not ideal practice to try and draw straight from your mind, especially if you are attempting to capture realism or improve your anatomical art skills – in order to ensure they can replicate a subject in all its beauty. Therefore, Midjourney, only operating off ‘prompts’ or words struggles to do this. You can see it very clearly when you request for (as the community dubs) ‘unreal engine 8’ style human features.

Here are examples of some images I created with my free trial, using prompts based around Alice in Wonderland by Lewis Caroll (such as roses; a vorpal blade) in a gothic reconstruction. Personally, I believe they are incredibly beautiful and evoke feelings of wonder as the others do. I upscaled them to the maximum, and although they are not horrifically unidentifiable, they replicate the idea of ‘uncanny valley’. The proportions aren’t built in the image of a real person, so they’re very abstract. No hand is holding the knife; rather it’s like an extension of the body that is doing so. In any instance, I want to highlight the brilliance of the platform, its utility for individual use, and the speed and talent of the bot, particularly as it was born from an independent small team that is seeking to innovate.

Applications of AI Art Platforms

Moreover, AI has become an extremely capable outlet for professional services. Playground AI can help with the creation of art, social media posts, presentations, posters, videos, logos and so much more. The opportunities are genuinely endless. Anyone who scrolls through LinkedIn every day will see their feed littered with generic and recyclable anecdotes which strangely generate a lot of buzz and interactions. The AI, too, has noticed this, and with a short prompt, you can invent a person and their career scenario with ease. It may certainly be able to write your essays for you, and no longer need to pay an external writer to help you with your university work! (but I didn’t tell you that). Irrespective of the abundance of creative opportunities and utilities that AI brings to the table, there are still some applications I disagree with.

The Repulsive Side of AI Art

In order to underlie my argument, I will present a case that is replicated over and over within many AI platforms, with some tools made solely for the purpose of stealing. What’s most infuriating is that this idea is illustrated as an innocent process where you are just providing a reference for the AI, so it can style it as you want.   

Look at Playground AI. This platform exercises a very similar process with prompts, a choice of up to 4 images, and options to increase image quality. It uses two models, Stable Diffusion and DALL-E 2, which organically generate digital images from natural language descriptions. Nevertheless, PlaygroundAI also has a function called ‘Image to Image’ and this is where it begins to spiral.

1: Prompts versus Image ‘Inspiration’

Inspiration? ‘Upload an image to use as inspiration?’ That’s not quite right. Somewhere the word ‘inspiration’ has replaced ‘stealing’, as art, usually created by independent, hard-working and self-employed artists, is unwillingly submitted and used by others with no effort or credit whatsoever, and this platform gleefully facilitates it.

Now, Suhail might challenge me on this, claiming it’s used for ‘tweaking’; interestingly, it’s those limitations that AI has, that was previously mentioned, that promote this. AI itself cannot generate a ‘perfect shot’, really emphasising the artificial aspect of its intelligence. It’s a bot, not an artist with a vision.

What’s even worse is the impact it has on artists.

I wonder what artists hate more – AI-generated art or AI NFTs?

No one can deny the empowering feeling of achieving something you’ve always wanted or being blessed with immense creative ability and realising something within your mind but being unable to materialise it for the world to see. Garry Tan’s perspective that ‘it’s addicting to have that much visual creative power’ exemplifies this. However, I would argue that it’s not necessarily ‘creative’. At best all you need to do is conjure up some words for prompts, but, most certainly, there’s this addiction to ‘power’. You no longer need to practice drawing for years, commissioning art, or search for hours on the web to find an art piece that fits the image in your head.

But what about artists? Not only is their art vulnerable to being stolen, but AI is rapidly disputing their jobs. Even more so, many are using AI art to profit from them, as you can see widespread adoption in the community of AI to create NFTs – digital information that cannot be exchanged or substituted for another, which has unique and distinguishable attributes. The concept of NFTS to me was always flawed, for anything on the internet can be leaked, replicated, or illegally distributed. Now, with the accessibility of AI which can manifest images in under sixty seconds, it is difficult for me to understand why any of it is valuable. In addition, the ‘uniqueness’ and ‘non-replicability’ elements are easily bypassed, for example, by using them as ‘inspiration’. Gatekeeping art becomes impossible.

This community of AI art is just one of many, however: many of which are gross and disturbing.

Strange Communities: What is Waifu Diffusion?  

Lastly, I would like to briefly refer to a trend called ‘waifu diffusion’, a hashtag or community name for people who use Stable Diffusion to produce, entirely, a woman with large breasts. It’s also not uncommon to find ‘huge breasts’ as a prompt. I hear there is now an AI tool to edit a woman’s pictures to make them appear nude, and it’s occurring a little too frequently.  

I fear the repercussions of the public, free-to-use, and costless AI art productions, and the platform they provide… But hey, if anything, at least you can look at AI art that’s generated by Taylor Swift lyrics.

Annie Phillips

+ posts

This author does not have any more posts.