The fight against AI art

[Jolene – Holly Herndon Version]

The song you’re listening to is a cover of Dolly Parton’s Jolene. Except this isn’t a cover by a human artist and there is nobody standing in front of a microphone…

The voices you hear were created by Artificial Intelligence.

Just like this isn’t actually the voice of multi-award winning actor Morgan Freeman: 

“If I said I wasn’t real, would you believe me?”

Diep Nep Youtube

This is a digital mimic too.

They’re part of a new artistic movement: AI-generated art. And it’s everywhere: AI music, AI paintings and even a new AI TV series. 

But one area in particular has created a lot of controversy: AI-generated imagery.

This is where computer software conjures up brand new images in response to text descriptions. A website called DALL-E is the most famous. 

So how does it all work? Here’s Joanna Stern from the Wall Street Journal: 

“The artificial intelligence is learning by looking at millions and millions of flash cards and what’s on those flashcards well it’s images and then it’s labels so it’s a picture of something in let’s say a monkey and it looks at thousands of pictures of monkeys that are labelled monkey and so the AI starts to learn that is what a monkey is. Then through all of this deep learning and all of this technology when you put in the phrase a monkey recording a podcast the artificial intelligence starts to build that image pixel by pixel and it’s not putting together images it’s already looked at it’s building that image from scratch.”

Joanna Stern

The internet is full of people experimenting with programmes like DALL-E, and the results are incredible. People without an artistic bone in their body can create beautiful, creative images of anything in any style, with only a few carefully chosen words.

You’re about to hear a woman who was sceptical about the power of AI-art at first, but she quickly changed her tune after she was shown the images the software had created…

“Okay, so a dark, spooky mythical creature inside of ancient ruins rain forest. I think there’s still nothing that needs an artist putting oil and minerals to a canvas.”

“The image that I’m about to show you right now is made in nine seconds no human being in the world ever would be able to make this image right here in seconds.”

“Oh my gosh, oh my god and there’s actually more images. Oh my god. Wow they made this at night, all four of these in nine seconds. I love it.”

iamLucid Youtube

But these AI-generated images don’t appear out of thin air. The AI systems have been ‘trained’ using billions of existing images – many of which have been created by real human artists. And this process almost always happens without the artists’ knowledge or consent. 

So the question is: how much of the new artwork is down to the original artist and how much is actually new in any meaningful way? It is a question which is beginning to create real legal headaches…

***

 “I thought that it was using only a photos to create some Landscapes and I didn’t think about you know using living artists as a as a style reference or style guideline for the the AI”

Greg Rutowsk

Greg Rutowski is a digital artist from Poland. He creates epic fantasy landscapes with dragons, wizards and one-eyed orcs. He has drawn illustrations for games like Dungeons & Dragons, and Magic: The Gathering. 

He’s also one of the most commonly used prompts in AI art generation. His fans type in things like: ‘Witch with sword in the style of Greg Rutkowski.’ 

His name has been used as a prompt more than 250,000 times so far. That’s over 100 times more than Leonardo da Vinci, Picasso and Michelangelo combined. 

That may sound exciting but for Greg, it has been a nightmare.

The moment an artist uploads their work online, it is absorbed by the AI software. And the impact on Greg’s livelihood has been devastating…

“I was really concerning about my future because we as a as digital artists we kind of like live in the in Internet and we kind of use different sites websites as a you know galleries. Basically social media and we our career is based on internet and then when you type in your name it should be like after many years of working it should be mostly focused on your output on your images and within like two months the Google results were yeah you know totally focused on the AI uh instead of me.”

Greg Rutowski

An artist might have spent decades dedicated to developing a distinctive style, only for a robot to copy it and create similar work in seconds. All without any compensation or credit. 

But now there’s a growing counter-movement which is fighting back.

***

Tensions over this issue are high in the art world. 

“You’re basically ripping someone else’s style off in the most low effort and no skill manner or manageable and then putting it forward as your own.”

GOTH ROSS Youtube

AI-generated work is so new that the legal ramifications and copyright – or ownership – issues are unclear. The art community was blindsided at first, but now it is making huge efforts to regain control. 

A new tool allows artists to attach a digital tag to their images with a code like ‘No AI’ which effectively hides it from the artwork-gathering algorithms. Others are coordinating a default opt-in system so only those artists who want their images to train AI systems will be used.

But that might not help artists like Greg whose work is already caught up in all this. At the moment Artificial Intelligence cannot erase anything from its memory, so it can’t forget all the tens of thousands of images it’s already been trained on. 

The reality is: AI art is here to stay. But that doesn’t necessarily mean human creators are doomed to be overtaken. Heavyweight players are wading in. The stock photo website Getty Images is suing one of the AI companies for breach of copyright. The fightback has begun. 

This episode was written and mixed by Rebecca Moore.