Is AI leading its own fine art – ificial movement? As the era-defining battle of man vs. machine takes centre-stage of the global culture wars, more major questions emerge: Can computers be creative? Who owns AI-made artwork? And should it be revered or decried as fraudulent?
February 27, 2023
It took Monet nearly three decades to complete his full collection of “Nymphéas” or “Water Lilies” paintings, yet images of the same Normandy water garden could be captured by a photographer in mere minutes – with arguably less effort.
You can imagine it now, as pointilism is replaced with point-and-shoot, 19th century landscape artists pose the question: Can cameras really be creative?
But, lo and behold, photography is still alive. Having earned it’s rightful place in the world of modern art, the pictures a camera captures are equally as revered as the oil-painted masterpieces of Monet and his contemporaries. Despite this however, history appears to be repeating itself.
Enter artificial intelligence (AI).
Although a lot has changed since Monet’s time, some things never do: people don’t like change, to be challenged or the unknown. AI is the embodiment and catalyst of all three.
What started with Arnold Schwarzenegger as a sunglasses-clad cyborg assassin, has now evolved into a full-blown era-defining culture war of man vs. machine, with many of the world’s artists heading up the frontline.
“AI is destroying fine art and undermining creativity,” say the artists.
“Computers fuel new levels of creativity,” says AI.
Amidst this battle, however, there is perhaps at least one thing that artists and AI can still agree on: Human hands are really hard to draw.
First, it came for the creatives
The fear of being replaced by machines is not a new narrative. Since the early days of automation we’ve known of the threat AI poses on various professions, with the blue collar workers first in line to be usurped, then the white collars, followed by the creatives right at the back of queue.
But the thing is, it’s actually gone in the exact opposite order.
As AI image generator platforms such as OpenAI’s DALL-E, Midjourney and Stable Diffusionhave erupted, both traditional and digital artists have found themselves going head-to-head with computers.
Many traditionalist connoisseurs have begun questioning if the images generated by computers can actually constitute art, while others have labelled the trend as a fraudulent farce which infringes on copyright and undermines the craft they’ve spent many years perfecting.
Some artists have even begun adding a “Do not AI” label to their Twitter profiles in protest.
Despite this, however, AI-made artwork has made its way mainstream, and computers have begun swiping competition trophies out from underneath them.
The AI art scene is booming
Deep Fake, NFT Paris, and Frieze are just some of the exhibitions that have shone a spotlight on AI generated artwork, while another AI artist reimagined women as different countries.
PATRÓN have launched their own AI image generator for tequila fans to design the margarita of their dreams.
Architects are using AI art generators to design and visualise fantastical structures. Imagine futuristic curvy concrete facades, biomimetic floating stadiums, urban canopy gardens – you get the picture.
Publishing and print industries are also embracing AI, with news outlets like The Economistand many authors now using automation to design cover images.
![Sony PlayStation Store 110 - Sony, [Digital]](https://i0.wp.com/i5.walmartimages.com/asr/5d3e8d52-3525-4e75-b862-10394c67a31c.c3f21de9f6a2f65537248191772f43b3.png?w=1380&ssl=1)
Even the fashion world is successfully navigating the space between neural networks and New York Fashion Week, with digital designers such as The Fabricant creating their own code-couture collections for people to purchase and wear virtually. Think — your very own personal designer “instagram” wardrobe — without the real-life clothes, but with the real-life price tag; virtual clothing is selling for as much as $9500 in some cases.
Salvador Dali vs. Salvador DALL-E
The ongoing debate all boils down to one particularly controversial question: Can computers ever replicate the distinct je-ne-sais-quoi of humanness that’s required for creativity?
It’s a distinctly intangible, difficult to quantify, and deeply personal quality – a far cry from the replication, repetition, automation, and rigidity we associate with computers – and it therefore feels alien to imagine an inanimate machine being able to achieve it.
Where there’s no feeling, there’s no imagination, right?
Marcus du Sautoy, author of “The Creativity Code,” would argue that AI actually helps to inspire creativity, as would historian Yuval Noah Hurari, author of “Sapiens” and “Homo Deus” in which he says that humans are an “assemblage of organic algorithms shaped by natural selection over millions of years of evolution,” and has been very vocal in his beliefs on the potential to replicate our in vivo humanness, in silico.
However, for most of us, AI remains a distinct square that doesn’t quite fit in with the niche fluid curves of Art or the arts.
A manifestation of this AI aversion was witnessed recently, when the Mauritshuis art museum in The Hague sparked controversy by loaning their original copy of Johannes Vermeer’s famous “Girl with a Pearl Earring” to the Rijksmuseum in Amsterdam, and in its place hung “A Girl With Glowing Earrings” by Julian van Dieken – an AI piece made using Midjourney.
Similarly, in Australia, an AI-made “photograph” of surfers on a beach fooled judges to win first place in a photography competition, only to later have the title removed once the panel found out the image was in fact computer-generated.
Innate or otherwise, as a global society we seem to have a deep-rooted dislike of things that challenge the status quo, but let’s face it, Picasso and Braque’s abstract Cubism was initially regarded as absurd rather than admired, yet their revolutionary ideas caught on eventually.
Is it possible therefore, that the AI art wave we are currently riding is just the beginning of a new artistic movement, just as Impressionism or Cubism were in their own times respectively?
Regardless of the answer to this question, and existential and cultural issues aside, there are a myriad of other legal and economic questions that also need to be addressed, namely: Who owns the artwork made by AI? And does the creation of it, as well as the system’s training with rights reserved images, infringe on copyright?
Who owns AI-made art?
In order to generate an AI image, the model first needs to know what you’re looking for – the internet is a big place without any guidance parameters.
This is where prompts come in. These short strings of keywords are composed of unique combinations of ingredients such as style, colour, mood and even existing artists’ names to help define the nuances of the intended resultant image.
For example, the following images were created by inputing this prompt into DALL-E “a futuristic version of Claude Monet’s ‘Bridge Over a Pond of Waterlilies’”:
Similarly, in his installation, “Memories of Passersby,” Mario Klingeman uses a neural network trained on a large database of existing classical-style portraits to generate unique and novel AI portraits in real-time.
Including the names of artists in prompts, or training models on existing artwork in this way is largely intended as a direction for the AI, to show it from where to draw inspiration.
The problem is, it also raises a whole host of consent and copyright issues, as anything from an individual’s unique brush stroke style can be replicated without any authorisation from the artist themselves.
For example, Greg Rutkowski, a digital fantasy landscape artist who has created game artwork for the likes of Sony and Ubisoft, is a more popular ingredient than Michelangelo, Pablo Picasso or Leonardo da Vinci in user prompts – popularity which comes with a hefty price, as the internet is now flooded with faux Rutkowski renders, but the real Rutkowski gets no recognition or credit for any of them.
At what point does inspiration become plagiarism?
Does the AI artist own the generated artwork?
There is a general consensus amongst both EU and US officials that the individual responsible for crafting a prompt and the resultant image it yields, is also the holder of that images rights.
However, this week the US Copyright Office moved to retract the rights to images in a graphic novel, “Zarya of the Dawn,” which author Kristina Kashtanova created using Midjourney.
Though Kashtanova is permitted to hold the rights to other elements of the book, such as the story, the office deemed the book’s images as “not the product of human authorship,” and therefore cannot be copyrighted.
So if the AI artist doesn’t own the artwork…
Do the original artists who inspired the artwork own it?
When an AI model scrapes the internet and takes inspiration from existing artwork or copyrighted images, shouldn’t the artists or image banks who created or own the originals have the right to lay claim, at least in part to the AI derivatives?
In this vein, Getty Images has announced it would be filing charges against Stability AI, Stable Diffusion’s parent company, for scraping Getty copyrighted images from the internet and including them in AI models.
Some artists have also initiated class-action lawsuits against platforms like Stable Diffusionand Midjourney which they allege have used their work within AI datasets unauthorised. In some cases artists signatures have even been spotted within AI generated images, further fuelling the creative community’s concerns over consent.
As a result, many artists are now calling for their work to be removed from the training databases of AI research institutes and companies, requesting that prior authorisation must be given before they are included in the first place.
Infringing on copyright laws in this way is something that is also often seen within the NFTsphere, where many artists have their work ripped-off and resold as counterfeit NFTs without consent. But in the case of AI generative art where the output is largely unique, many argue that neural networks are simply drawing inspiration from the artwork they observe, much like any artist would – computers are just much faster at it than we are.
In the traditional sense, artwork is borne from the accumulation of observations, experiences and memories that an individual collects throughout their life – kind of like a subconscious bookshelf – which ultimately culminates in inspiration and expression. Does artwork in the AI sense not work the same?
We perhaps therefore need to ask another question: Who owns inspiration?
In attempting to define who owns what, at the level of what a computer has “seen” or taken insight from, aren’t we effectively putting a price-tag on intangibles such as nuance or style, which cannot really be objectively owned?
A lot of questions, for which there are not many concrete answers at present unfortunately, so we’ll have to decide for ourselves. But if the both the artist creating the AI-made artwork as well as the artists whose work inspired it don’t own the generated content, dare we ask…
Does AI itself own the artwork it produces?
Given that the entire world still seems fairly split on the question of whether AI possesses creativity even at this point, it’s probably a stretch to conclude that it will be granted rights to its own artwork in the near future.
Besides, we can probably all agree (or at least hope) that AI doesn’t possess the level of autonomy required to own things… or does it?
AI can already do so much; write essays, articles and poetry, answer the world’s biggest questions, compose music, give legal advice, diagnose illness, and help with national security, but as OpenAI CTO, Mira Murati warns:
“Very quickly we can end up in a place where machines are far more capable than us.”
“As AI systems get more capable, they don’t automatically become better at doing what humans want,” says OpenAI’s Human Alignment team lead, Jan Leike, “In fact, sometimes they become less inclined to follow intentions.”
For this reason, it’s of critical importance that in parallel with pushing the human-like advancement of AI systems, we must also ensure their development is aligned with human values, lest we risk a rapid loss of control.
“Solving this problem is of critical importance if we want life on Earth to go well,” says Murati.
AI creates brand new forms of expression
Yes, true, but it also perpetuates age-old systemic forms of bias.
The algorithms that AI image generation platforms run on are trained on billions of existing images tagged with keywords which supposedly describe them. Unsurprisingly, given the stereotypes entrenched within the infrastructures of society, when making connections between these descriptions and their visual representation, AI is not immune to bias – in fact, as it learns, it perpetuates the issue exponentially.
As author of “Man-Made,” a new book about “how the bias of the past is being built into the future,” said in an op-ed for the Guardian, “These baby biases become troublesome teenagers through machine learning. The bots are increasingly bigoted, like white supremacists neck-deep in conspiracy theory websites.”
However, if initiatives such as OpenAI’s Human Alignment project are successful, we can hopefully ensure AI develops safely, sustainably, and in a way that brings benefit to humanity – as is in fact OpenAI’s avowed mission.
In the same way that we prompt AI to create images, it can prompt us to discover new levels of creativity in return.
Rather than replace anyone, perhaps AI can be considered as a tool that provides an alternative angle from which to observe and create, exponentially increasing self-expression in new and exciting ways as part of a movement of machine-augmented creativity.
Kind of like the latest hi-tech model of paintbrush?
Subscribe to our newsletter.
This article was originally published on IMPAKTER. Read the original article.