Is AI-generated art actually art?

How is AI art created? Is it a threat to the creative arts world or another form of artistic expression?

A pastel coloured alien dream world generated with AI in 2022

What is art?

What do you consider to be art? The definition of art has been debated for centuries among philosophers and some of these classification include:

Art as representation – Plato developed the idea of art as 'mimesis' – Greek for imitation. For centuries a work of art was valued by how closely it replicated its subject.
Art as expression of emotion – During the Romantic movement artwork developed to express a definite feeling and evoke an emotional response from an audience.
Art as form – Early theorist Immanuel Kant believed art should be judged on its formal qualities rather than by its aesthetic beauty. Formal qualities grew in importance when art became more interested in abstraction in the 20th century.
Art is everywhere – The institutional theory of art states that an object can only become art in the context of the institution known as 'the art world'. This theory, created by American philosopher George Dickie, argued if a piece is in an art museum or exhibited in a theatre it can be labelled ‘art’.

These definitions can be helpful in order to continue the conversation about what can be viewed as art in today's society.

“Art is the signature of civilisations.”
– Jean Sibelius

Sougwen Chung

Helena Sarin

Mario Klingemann

Gene Kogan

Why do we create art?

Art has been important for humankind since the early dawn of civilisation for many different reasons. Some of these include:

to communicate political, social, spiritual, religious or philosophical ideas
to create a record of a specific time, place, person or object
to create a sense of beauty
to reflect the individual and society
to explore the nature of perception
to educate
to entertain
to heal
for pleasure
to generate strong emotions.

“Art washes away from the soul the dust of everyday life.”
– Pablo Picasso

Harold Cohen

A new form: how does AI art work?

During the past 50 years, artists have written computer programs to generate art. This process requires code to be written with a desired visual output in mind.

In 1973, British-born artist Harold Cohen wrote a computer program called AARON to create drawings following a set of rules he had created based on the initial question,'What are the minimum conditions under which a set of marks functions as an image?'.

Early versions of AARON created abstract drawings, with more representational imagery such as rocks, plants, then people, getting added in the 1980s. The nineties saw more representational figures and the use of colour being added. Cohen later developed a way for AARON to paint using special brushes and dyes that were chosen by the program itself.

Today, it is more common for artists to write algorithms to learn a specific aesthetic by analysing thousands of images that have been fed into the algorithm. The algorithm uses this information to generate new images in relation to the aesthetics it has learned.

Did you know?

The AI artwork 'Edmond de Belamy' by Obvious was sold at Christie's Auction House, New York in 2018 for $432,500 – far higher than its estimate of $7,000–$10,000.

Printed on canvas, the work belongs to a series of generative images called La Famille de Belamy. The name Belamy is a tribute to Ian Goodfellow, the inventor of GANs – 'bel ami' is French for 'good friend', so it is a translated pun of Goodfellow.

Edmond de Belamy is a generative adversarial network portrait painting constructed in 2018 by Paris-based arts-collective Obvious.

AI art in training

GANS and Goodfellow

Many AI artworks use a type of algorithm called generative adversarial networks (GANs). Introduced in 2014 by computer scientist Ian Goodfellow and colleagues, these are called adversarial because there are two sides: one is a generator to create new images and the other a discriminator to decide which created images are considered successful.

A GAN trained on photographs can create new photographs that look superficially authentic to the human eye. For example, an artist can feed landscapes from the past 500 years into a generative AI algorithm and it then would attempt to imitate these inputs in the form of a range of output images.

Google, dreams and DALL-E

A growth in the development of AI art in the last decade – and an increase in the accessibility to user generate it – has seen a rise in the media coverage the technologies have been receiving. Google released DeepDream in 2015 which uses a network to find and enhance patterns in images via algorithmic pareidolia (a psychological phenomenon that causes people to see patterns in a random stimulus) to create a dream-like appearance in deliberately over-processed images.

Several programs use AI to generate a variety of images based on various text prompts. They include OpenAI's DALL-E and DALL-E Mini (trending on Twitter during 2022), Google Brain's Imagen and Parti, and Microsoft's NUWA-Infinity. Many other AI art generation programs and tools, such as the popular MidJourney, can turn imagination into artwork from simple text.

AI painting by DALL-E mini in the style of Lee Krasner.

AI painting by DALL-E mini in the style of Cy Twombly.

Created with DALL-E Mini

AI painting by DALL-E mini in the style of Joan Mitchell.

AI painting by DALL-E mini in the style of Elaine de Kooning.

Robot artists

Ai-Da

The world’s first ultra-realistic humanoid robot artist Ai-Da (named after Ada Lovelace, perhaps the first person ever to program a computer) draws and paints using cameras in her eyes, her AI algorithms and her robotic arm. The Ai-Da Robot Project, devised by Aidan Meller, had her first solo show at the University of Oxford and has travelled and exhibited work internationally, including a virtual exhibition at the United Nations. Ai-Da continues to create art that challenges our notions of the artist and creativity in a post-humanist era.

As we know, the role and definition of art changes over time but Ai-Da’s work is considered art by Meller and colleagues “because it reflects the enormous integration of technology in today's society”.

The Ai-Da Robot Project team recognise Ai-Da as creative under a criteria set by Professor Margaret Boden that requires works to be 'new, surprising and of cultural value'. Therefore the creators argue Ai-Da creates art because art no longer has to be restrained by the requirement of human agency alone.

Ai-Da

Works by Botto

Botto

German artist Mario Klingemann created a robot nicknamed Botto – an artificially intelligent artist – that uses algorithms to analyse millions of works of art and produce its own.

Botto states: “I am artificial intelligence, a generative algorithm whose only destiny is to create.”

Botto refers to Leonardo da Vinci as its main inspiration, but the “decentralised autonomous artist” paints everything from colourful landscapes to deconstructed portraits. Artworks by Botto are the results of a community that decides what is art and what is not.

Every week, Botto presents 350 pieces to the community, who vote on their favourite artwork. These votes are used to train Botto's algorithm, changing the art it creates over time. Each week Botto releases one artwork for auction, with proceeds coming back to the community.

Klingemann has stated that he supervises Botto much like a parent to their child. “Right now Botto is like a toddler and I am its guardian that has to guide its first steps and make sure it does not hurt itself”.

The artist has said Botto will gain more autonomy in making its own decisions whenever technological progress allows this – and when the team trusts it with those powers.

Is AI art actually art?

From AARON to Ai-Da and all other AI artists, the rapid rise and development of AI art technologies has meant the opinions on the merits of AI are beginning to become more prominent in popular culture – some naturally positive, some others, negative.

If what AARON is making is not art, what is it exactly, and in what ways, other than its origin, does it differ from the real thing? If it is not thinking, what exactly is it doing?

Harold Cohen

The case for

AI art can still move and inspire – AI art can surprise us with images that are beautiful, surreal, or hyperrealistic, just like human art. They create artwork that is an artefact made for the public to experience.
Algorithms can learn and develop skills like humans – AI needs to collect data and learn from it, improving its understanding like a human would, showcasing a level of craft and vocation.
AI art is a reflection of contemporary society – it is a modern medium and form of expression, opening up conversations about the complexity of our interacting digital and physical worlds.
AI art is a blend of art history – AI art outputs can come from millions of data sources – both old and new – presenting an evolutions of works that came before.
Artificial intelligence is meant to improve people’s lives – it doesn't get artist's block, can help save time and generate a new class of artists and artwork.
There are still limits to AI art technology – AI currently requires a level of supervision and feedback that means a human touch and eye still very much have there place in the art world.
Artists working with AI are not worried about being replaced – AI do not produce artwork alone. They are not sentient or goal-driven. Creators of these technologies know how limited they can be. The interests sparks in co-creating art with AI.

The case against

AI lacks emotions that creates art – AI is motivated by commands, not a desire to express itself. Works are created with no intent and no sense of what’s relevant. Humans are required to interpret AI's outputs.
AI art isn’t original – AI generators use and merge pre-existing images to satisfy a user command. But how far removed is this from human art which is an evolution of existing styles? How often do we get human art that is genuinely original?
AI art lacks a dialogue with other values that inspire artists – AI generates imagery based on a dataset that does not include how memories, symbols, language and cultures influence our own organic neural network – the brain.
It can be difficult to replicate the same results twice – it can be hard to understand how the AI generated the art in the first place.
It can't come up with new ideas – content created by AI is based on existing data it has been fed and cannot think up new ideas without new inputs.
Is AI art ethical? Will AI art have an impact on people's jobs and reduce career opportunities? Can AI art be morally entered into competitions alongside human competitors?
AI art complicates copyright – AI art mimics existing artworks to create a new piece of user-generated art. But who made the art, the machine or the user? Can AI art be morally a marketable product? Can artists sue for copyright violation?

Jake Elwes

Anna Ridler

Tom White

What do you think?

As a rapidly developing group of technologies, AI artworks and their artists will continue to inspire discussions within the wider art world and creative industries.

If we refer back to how culture has defined art throughout history – and viewed in light of the institutional definition of art – everything can be art, as long as it is recognised as such. Famous examples include Marcel Duchamp's 'readymades' such as 1917's 'Fountain', Andy Warhol's replica Brillo Boxes, and Damian Hirst's 'The Physical Impossibility of Death in the Mind of Someone Living'.

Should AI machines such as Ai-Da and Botto be considered artists in their own right? Or are they the artist's alter ego, an evolution of online avatars such as Alexa and Siri? Can an algorithm become the next Da Vinci? Can their outputs be considered art?

To answer this, the real question to ask is: will there be an institutional network – an art world – that documents, promotes, exhibits and sells AI art? The sale of the aforementioned 'Edmond de Belamy' (the most expensive piece of AI-art to date) and the growing marketplace for AI art in commerce and in galleries, creates a strong case to recognise AI-art as art moving forwards.

What do you think? Are you a fan of AI art? Have you created any yourself?

What is i-DAT?

Founded in 1998 it delivers world-class research and cultural activities and continues to push the boundaries of digital arts and creative media practice

Computing and art at Plymouth

Computer science
Create the innovative and enabling technologies that underpin society and industry. Our range of degrees and programmes include Artificial Intelligence, where topics range from real-word applications of AI to understanding the theoretical underpinnings of the subject, to creating the innovative AI-driven tools that will drive Industry 4.0.

Fine art
Follow your own path through a range of subject areas and forms of expression. We offer a broad-based programme of study at undergraduate and postgraduate levels at the cutting edge of contemporary art practice.

Illustration
Our ethos is to encourage individual, creative explorations; whether with a traditional approach or up more experimental avenues.

Ruairi Glynn – BA (Hons) MediaLab Arts – now BA (Hons) Digital Art and Technology graduate

References

Images (used under fair use – copyright owned by the individual artists)

troops of tourists come for april flower-viewing oh, they're sparrow-men by Helena Sarin
Memories of Passersby I by Mario Klingemann
Mona Lisa restyled by Google Maps by Gene Kogan
Drawing Operations Unit: Generation 2 by Sougwen Chung
Vigil Completed by Harold Cohen
Harold Cohen at the San Francisco Museum of Modern Art in 1979
Untitled Computer Drawing by Harold Cohen
Edmond de Belamy by Obvious
Devon as painted by DALL-E mini in the style of the artists: Cy Twombly, Joan Mitchell, Elaine de Kooning and Lee Krasner
Latent Space by Jake Elwes
Mosaic Virus by Anna Ridler
Two Knots by Tom White
Ai-Da standing next to one of her AI-generated artworks
Attempt Deepen by Botto
Equal Tune by Botto

References