AI Artist Stefan Olivier’s Anthozoa Specimen series
Step into the fascinating realm of AI art with us. Artist Stefan Olivier is exploring the limits of AI-generative art with his mesmerising creations. It is our pleasure to be showcasing four of his pieces during the Into the Deep group exhibition on display this June.
We got talking to him to find out about his creative process, the role of AI in his work and the captivating possibilities that emerge when man and machine collaborate. Read on to gain a deeper understanding of the intriguing realm of AI art through the eyes of Stefan Olivier.
How would you describe AI art to someone who is unfamiliar with the concept?
Think about it as painting with a mechanical brush that has its own interpretation. You don’t have full control but you have to work with the mechanical brush to bring your ideas to life. Two artists working together, inspiring each other, but also fighting to achieve their respective visions. There are many failures, and many different ways of tackling the challenge, and only after many attempts together can you achieve something that is not 100% what you envisioned but hopefully something you have co-created that is on par or even better than your original vision.
“The prompt engineer has to rework the prompts and manipulate them to eventually get to a pleasing or accurate visual that makes sense to a human.”
What inspired you to explore AI as a medium for artistic expression?
I have a career as a digital designer, creator, and advertiser in the digital realm with a graphic design and fine arts education background. Ever since we got our first family computer I have been captivated by what I can create, or artificial worlds I can get lost in. To be given a tool that can bring my concepts and ideas to life in such an interesting way was a natural progression for me.
Can you explain the process behind creating your mandala-like designs using underwater organisms in particular?
Originally I started creating mandala designs, but it very quickly became obvious to me that I was only limited by my imagination and that mandalas don’t have to conform to any specific rules or mediums. In my career as a conceptual advertiser, we learn that the best brainstorming ideas come from breaking all the rules and that there are no bad or weird ideas. This inspired me to break the rules and I started experimenting with creating mandalas inspired by or consisting of anything in the existing universe. As I’m often inspired by the natural world around me, manipulating this in a digital world opened up so many new possibilities. As mandalas are incredibly geometric there was a natural commonality between controlled or designed geometry and the geometry in the natural world, like the reassurance of the Fibonacci ratio. Furthermore, the human brain is designed to recognise patterns and is naturally drawn to geometry, as it brings a sense of order in a chaotic world. My intention was always to create something that makes a statement wherever the artwork may be displayed, to draw the viewer’s eye. By emphasising this natural geometry I could play on these built-in triggers that draw the viewers into the artwork, just like mandala designs do. The corals are one of many subjects that I am exploring with this concept.
How does the AI algorithm you work with interpret and generate art based on these organisms?
Generative AI models work on the concept of learning with large data sets. It knows what corals are from viewing millions of images that it has been told are corals, likewise, it knows what geometry is, and it knows what photographs are, 3D visualisations, or paintings. It has millions of references that it has studied to understand the characteristics that make a coral a coral, and the same for any other subject. Where the magic happens is how it works with the prompt engineer or artist to combine these elements into a single visual. For the AI it is as easy to create a cat sitting on a matt as it is to create a cat made from cheese. Whether this looks real or pleasing to a human, however, it does not know.
What role do you play as the artist in the collaboration between yourself and the AI?
The prompt engineer has to rework the prompts and manipulate them to eventually get to a pleasing or accurate visual that makes sense to a human. When you prompt it, it will give you a few options, you can either ask it to try again, or you can rework your prompt or you can pick an option and rework that specific option. It’s like breeding with dogs over time to get a specific new breed of dog with its own desired look and features achieved. The real creativity comes from the concept the artist imagines, and then you work to try and achieve this vision over time. Constantly manipulating it until you get the desired effect.
How do you strike a balance between the creative input of the AI and your own artistic vision?
This is where the concept behind the work needs to shine through. If the artist owns the concept and it is strong then the AI is simply aiding you in making it come to life, rather than driving it. I could ask it to paint a cat in the style of Picasso and the result could be quite pleasing but have I, the artist, really contributed anything at all? On the other hand, if I asked it to imagine that rabbits somehow became the dominant species over the next thousand years and had conquered our solar system as a space fairing race and then asked it to illustrate specific scenarios depicting everyday life in a certain style, then I am most certainly driving the concept as the artist and the AI is a tool or medium that is helping me realise my vision.
Are there any specific challenges or limitations you encounter when working with AI art?
Yes, there are. You will never be able to replicate the exact clear vision you have in your head. The AI simply cannot see the vision in your mind, so you have to relinquish the fact that you are co-creating. You may in certain instances achieve very close to your vision, but never exactly. It can also be very difficult to create a series. As mentioned before, you will create hundreds of versions in your pursuit to achieve your desired result, now you have to create a series that has the same feeling and share certain qualities. This almost has to happen simultaneously as you are creating and different version start to branch off, you need to try and keep them all on the concept as they develop together. It is very difficult to fully control the AI completely, but through practice and mastering the art, your can control this tool more and more, just like any other medium.
How do you see AI art influencing or transforming the traditional art scene?
Firstly I believe that traditional hand-crafted art will grow in value as we move into the exponential age. I believe you will see creativity grow all around us because the ability to express ideas or concepts is now accessible to everyone. When more people are contributing to creative ideas and expressing them they are inspiring new ideas in others around them all the time. You may notice some traditional artists’ work shifting over the next 5 years as they use ai to conceptualise ideas that they then bring into their traditional art, using it as part of the creative process.
What do you hope viewers will take away from experiencing your AI-generated artworks?
A better understanding of, as well as the value of AI as a medium. A sense of wonder and at the world of aesthetically pleasing possibilities that can exist in a frame.
In your opinion, what is the unique contribution that AI art can make to the broader artistic landscape?
99% of artists work in isolation on their art, but co-creating with AI pushes one past their own boundaries and can result in beautiful new outcomes. This along with the inspiration and aid that it can provide traditional artists in their own process is also significant.
Since AI models are learning from existing photos and graphic formats, is it not copyright infringement?
AI is learning from these images the same way you are learning from and inspired by other artists or existing imagery. Everything is already a remix of everything else. If this wasn’t the case we would have never had the Impressionist, Renaissance, or Surrealist movements because artists of those periods would merely be painting 100% from their own mind and not be inspired by the zeitgeist of the time. Generative AI models are merely mimicking the equivalent human experience.
Be sure to visit the gallery during the month of June to view Stefan’s works.