Contemporary Art

AI in Art: What Does It Mean?

Agnieszka Cichocka 15 December 2022 min Read

I keep wondering if Leonardo da Vinci, who, in my opinion, was the most forward-thinking artist of all time, would have ever imagined that art would one day be created by AI. He worked on numerous ideas and was constantly experimenting and, although some were failures, he persistently tried new products, helping to move our world forward. Without such people, progress would not be possible. 

Samsung AI lab and Skolkovo Institute of Science and Technology, Few-Shot Adversarial Learning, 2019.
Samsung AI lab and Skolkovo Institute of Science and Technology, Few-Shot Adversarial Learning, 2019. Synced.

AI animates paintings and photos without 3D modeling. Posted by kyus2001 via YouTube, 24th May 2019.

Da Vinci’s willingness to cross intellectual boundaries helped him solve problems in new ways. Moreover, he was never satisfied with the status quo and became a constant innovator. His courage in questioning existing assumptions and finding unique ways of thinking was legendary. He drew visions of planes, helicopters, parachutes, submarines, and cars more than 300 years before any of these ideas were improved.

Why am I writing so much about him? I believe he was a visionary who pushed the arts into new pathways. In the same way, art is being pushed right now by those who create new artistic genres technology, and science. They are exploring and experimenting with the art of the 21st century – art supported by AI or authored by robots. We often hear that in art there is nothing more that can surprise us. Are we so sure about that?

What Is AI in the Arts?

Pindar van Arman, Bonnie, 2020. AI art
Pindar van Arman, Bonnie, 2020. Morf Gallery.

Artificial Intelligence art piece refers to any artwork created with the assistance of AI. It can be a work created autonomously by AI systems or a work that is a collaboration between a human and an AI system. AI is an area of computer science focused on building machines that simulate human cognition, learning, and mimics. AI-generated art can include fully autonomous works—where researchers explore whether AI can be creative without human intervention to human—as well as human-robot collaboration in real-time. By doing so, they try to understand the communication between people and machines. Artificial Intelligence is now part of some paintings, music, poetry, and films.

MORF Gallery Presents – Pindar Van Arman. Posted by MORF Gallery via YouTube, 13th April 2021.

There are several subsets of applications of AI such as machine learning, deep learning, embodied AI, artificial neural network, generative adversarial network, creative adversarial network, etc. Seems like a lot of difficult words right? Let’s have a closer look at them to better understand how AI works in artworks and why this kind of creativity is becoming a new artistic genre.


Mikael Hvidtfeldt Christensen, Octopod, 2007. AI art
Mikael Hvidtfeldt Christensen, Octopod, 2007. Photo by Mikael Hvidtfeldt Christensen via Wikimedia Commons (CC BY 2.0).

In simple words, an algorithm is a sequence of step-by-step instructions. These might be very simple directions like  “turn left, turn right, go straight on.” Or they might be very complicated, like hundreds of lines of code with instructions that tell the computer what to do, how to make calculations, recurrent problems, and take other specified actions.

Machine Learning

As humans, we learn by acquiring knowledge through observations, senses, experiences, etc. This is similar to computers. Machine learning is a process in which a computer system learns how to perform a task better in two ways—either through exposure to environments that provide punishments and rewards (reinforcement learning) or by training with specific data sets (the system learns automatically and improves from previous experiences). Both methods help the systems improve their accuracy. Machines then use patterns and attempt to make an accurate analysis of things they have not seen before. To give an example, let’s say we feed the computer thousands of photos of a dog. Consequently, it can learn what a dog looks like based on those. Later, even when faced with a picture it has never seen before, it can tell that the photo shows a dog.

If you want to see some creative machine learning experiments in art, check out ML x ART. This is a website with hundreds of artworks created using AI tools.

Homepage view of MLxART
Homepage view of MLxART. Accessed on 25th August 2021. MLxART.

Artificial Neural Networks

Neurons are brain cells that are designed to transmit information to other nerve or muscle cells. A neural network is a computer algorithm that attempts to simulate the network of neurons which make up our brain. Neural networks include layers of connected “neurons” that send information to each other. In contrast to machine learning, neural networks learn what to do on their own without specified step-by-step instructions.

Deep Learning

Deep learning is a subfield of machine learning. While machine learning uses simple concepts and step-by-step guides, deep learning refers to neural networks with multiple layers of connected neurons. Most importantly, deep learning is designed to imitate the human brain.

Embodied AI

Embodied AI is a type of artificial intelligence that controls something in physical space –- a body, robot arm, etc.

Sougwen Chen, Why I draw with robots, September 2019 via TEDtalks.

Sougwen Chen is an internationally renowned artist and researcher. Her works are a combination of marks made by hand and marks made by a machine. She experiments with the understanding between humans and machines. She was a research fellow at MIT’s Media Lab and was selected as the Woman of Year 2019 in Monaco for her achievements in Arts & Science.

Sougwen Chen, Omnia per Omnia, 2018. AI art
Sougwen Chen, Omnia per Omnia, 2018. Artist’s website.

Generative Adversarial Network

GAN is a machine learning model where two neural networks (Discriminator and Generator) compete with each other to become more accurate in their predictions. What’s more, they progressively improve toward the desired goal.

The goal of the Generator is to generate “fake” images and the discriminator tries to spot “fake images” developed by the Generator with the help of a provided data set. First, one network tries to create a new output based on a training set. For example, it might create a portrait. Then, the other network tries to identify where the output is different from the original training set. In this way, one network gets constant feedback, getting better and better at creating an image that is closer to the desired outcome.

They create an output, judge its similarity to the initial data set, and repeat the process all over again. Finally, they come up with an outcome that is astonishingly lined up with the original training set but, on the other hand, is an entirely new image. At the end, the Discriminator won’t be able to tell the difference between generated imagines and images in training. This means that the goal of the Generator is accomplished.

Robbie Barat, Nude Portrait #1, 2018. AI art
Robbie Barat, Nude Portrait #1, 2018. Artnome.

Robbie Barat is a young artist who creates AI-generated nude portraits, landscapes, and sculptures. Using an implementation of GAN and thousands of examples of nude portraits or landscapes from WikiArt, a neural network was trained to create artworks in these genres. He completed his first AI art project when he was in high school. He taught a neural network to rap like Kanye West.

Creative Adversarial Network

In simple terms, CANs are GANs that can think creatively. The outcome is more random and often surprising as it attempts to pattern natural human creativity. Above all, the output can also be trained to know what type of creative quaintness is attractive and which is not.

Dr Ahmed Elgammel is a professor and researcher exploring how AI can be creative. He is the creator of AICAN, the first machine-generated artist to pass the Turing test. It is a test of the machine’s ability to act like a human so that you can’t see the difference between the computer and human behavior.

AICAN + Ahmed Elgammal, Faceless Portrait #2, 2019. AI art
AICAN + Ahmed Elgammal, Faceless Portrait #2, 2019. HG Contemporary.

In conclusion, the mixture of art and technology is a recent phenomenon and, in my opinion, our future. However, the tech language is not an easy one to understand, and that is why I aim to bridge these two disciplines and translate art into technology and technology into art.


Contemporary Art

Why You Should Visit the Boros Collection in Berlin

The Boros Collection, alternatively called the Bunker of Berlin, Banana Bunker, and, according to the New York Times, the “Hardest Club on...

Kaena Daeppen 4 December 2023

Contemporary Art

5 Japanese Contemporary Artists You Have to Know

Japanese artistic tradition spans centuries and has a far-reaching influence. Today, the artistic scene in Japan is more dynamic than ever, and these...

Carlotta Mazzoli 27 November 2023

Gallery view of the Marina Abramović exhibition at the Royal Academy of Arts, London, from 23 September 2023 – 1 January 2024, showing Four Crosses (detail), 2019. Courtesy of the Marina Abramović Archives. © Marina Abramović. Photo © Royal Academy of Arts, London / David Parry Contemporary Art

Marina Abramović at the Royal Academy in London

In 1997, Marina Abramović (b. 1946) attempted to scrub clean over 1,500 cows’ bones at the 47th Venice Biennale. The performance, titled Balkan...

Ania Kaczynska 9 November 2023

Contemporary Art

Indulge in Juicy! A New Exhibition by Florence Houston

Taking place at J/M Gallery, London, Florence Houston‘s upcoming exhibition Juicy! (1-7 November 2023) showcases delectable still lifes.

Isla Phillips-Ewen 26 October 2023