When Artificial Intelligence Creates Art

From a psychological perspective, humans create art for several reasons: to elicit an emotional response, to remember past events, to communicate, and to educate.

To cut a long story short, art is something that we create to understand who we are as individuals -to express the inexpressible and to communicate beyond words. We want to create experiences that trigger feelings and affect the perspectives of others.

Humans didn’t have time for this kind of soul searching up until 30 thousand years ago. We spent our whole lives hunting, fighting, staying warm, and seeking shelter. 

But then something happened. A new species of human discovered how to gather food. Cro-magnon tribes finally settled when they were no longer plagued by frequent hunting. They sat and watched their fires throw lengthy shadows on the walls of deep tunnels, thinking about something other than surviving for the first time.

They started drawing with burning sticks.


They couldn't have done it without the cognitive abilities that their predecessors had acquired through hunting: manufacturing and utilizing tools, developing memory, creating language, expressing themselves, and identifying patterns in the environment around them. 

Memory, language, cognition, reasoning, learning, expression, and pattern recognition are all qualities that we are currently attempting to replicate with machines. These are the fundamental elements of AI.

We initially taught AI to understand and imitate our work before utilizing it to create art.

Style transfer is a technology that uses deep neural networks to reproduce, recreate, and combine art styles. It takes artistic components from one image and integrates them into a new one. No prior artistic or coding knowledge is required. 

The premise is the same whether it's applied to art, photography, cinema, or music: pick a piece of art whose style you wish to imitate, then let the algorithm apply that style to another image. Choose different art styles and let the AI create mash-ups that gradually blend them.

You can see the four original pieces of artwork they chose in each corner in the Google AI sample below. Between these corners, a grid of photos represents the degrees of blending one image into another and applying the resulting style to a new shot. The central image was produced by combining all four elements in equal amounts.


Imitation is interesting and perhaps profitable, but it is not in the true spirit of art. It's merely reflecting what we've already spoken back to us. 

We need to use AI as more than a copy machine if we want it to help us say something new.

Art is required to envision what AI can become and to comprehend its impact on who we are becoming. So let’s take a moment to think!

What does it mean to have a connection with a machine? 

Will having access to the entire world's knowledge impact the way art communicates and connects? 

Could our partnership with AI lead to previously unimagined forms of art? 

Could it alter our perceptions of one another? Is it possible to communicate across borders? Even in the future? 

Is our culture evolving as a result of it? Is AI coming up with its own?

There's no arguing that by giving computers the same powers that motivated us to create art (memory, language, expression, empathy, reasoning, and learning), they might one day opt to create their work. 

Why would the first "AI caveman create art with a burnt stick" do it? 

What is it going to be attempting to figure out about itself?

Zeynep B. Gergin / Team Hub21