The German photomedia artist Boris Eldagsen won the Sony World Photography Awards in April 2023. The unusual thing: He entered with an AI-generated piece and ultimately rejected the prize. Eldagsen weighs in on AI and creativity.

In the last year, I have taught over 30 workshops, given 60 lectures, and taken part in 20 panel talks. I repeatedly come across unfounded fears that result from a lack of practical knowledge and/or well-defined terminology. "AI replaces human creativity," "AI makes artists superfluous," "AI is more intelligent than humans and is developing consciousness."
When I ask what creativity, intelligence, and consciousness are, there is silence. What a missed opportunity for a deeper debate! I studied philosophy for six years and learned to appreciate when people take the trouble to define terms. Definitions are the best basis for a discussion. Otherwise, it's all just opinion. And mostly one-dimensional nonsense that overlooks the complexity of the world.

All the terms used in the media and panel talks are "philosophical," ultimately not clearly definable terms that are subject to change. But we use creativity, intelligence, and consciousness as if they were unique concepts. You could spend several semesters at university studying each term without coming to a clear conclusion. Nevertheless, everyone thinks they know what is meant by "creativity" and the fear that AI could take it away from them.

So, let’s have a closer look:

What is creativity?

Wikipedia defines it as something new and useful, referencing the Creativity Research Journal, Volume 24: "Creativity is the ability to create something that is new or original as well as useful." Whether this is global, continental, national, community-based, or just refers to the development of a single person remains open to interpretation. Why creative things should also be useful is a mystery to me. Especially because I can attribute a use to most things.

Collaboration with AI: What role do humans play?

AI is an ideal tool for artists. It frees artists from material constraints and allows them to draw entirely from their imagination. The material they are working with is their own knowledge and experience. As the creative process is a co-production of humans and AI, the role of the artist is comparable to a director. The artist ceases to be a single voice and instead becomes the conductor of a gigantic choir consisting of training data.

Who has the lead in the creative process: man or machine?

Many critics of AI believe that working with AI means giving up one’s own creativity. They overlook the fact that the creative process consists of three steps: prompt, generation, evaluation. The more informed the prompt (the start of the generation process) is, the better the generated images will be. The more informed the evaluation of the generated images is, the easier it will be to adapt and improve the prompt in the next step. It is here, in the first and third steps, where the artist connects his subjectivity, his artistic career, with the training data to create a new work. These steps can and must be fine-tuned in many runs — until a convincing result is achieved.

Is human creativity replaceable?

Last autumn, many international studies were published that celebrated ChatGPT’s performance in creativity tests, which rivaled humans. But what definition of creativity were these tests based on? Unfortunately, this was not communicated.

Which creativity theory can explain what is going on?

Creation myths aside, the yield from 150 years of creativity theory is unfortunately meager, and more recent buzzwords such as "design thinking" have unfortunately been unable to add anything substantial. I would recommend exploring the ideas of Margaret Ann Boden (born 1936). Her work embraces the fields of artificial intelligence, psychology, philosophy, and cognitive and computer science.

She suggests a threefold division of creativity:

a) The combination of given content (ex: mashups such as Batman in Barbie style)

b) The exploration of a given space (we humans have a certain perspective in space, but AI can capture all perspectives, looking under the chair, behind, and above us).

Only a) and b) were researched in the studies — and they are naturally manageable for AIs that are based on statistical and mathematical probabilities.

But what about the third form of creativity according to Boden:

c) The transformation of given content and spaces (breaking the rules of content or space to transform it into something new)

Is this even possible? Can AI free itself from its training data and take a leap into another dimension?

At Munich Media Days (Europe’s largest media convention), I asked Professor Björn Ommer, whose team built the foundation of the open-source model Stable Diffusion (a text-to-image generative AI that creates digital images based on text prompts), about this. He is extremely skeptical that this is possible.

Thus, we must develop a new distribution of tasks between man and machine. What if we use AI for combinatorial and explorative creativity and concentrate on transformative creativity? After all, just as the quality of human chess has improved through training with invincible chess computers, human creativity could also grow in collaboration with AI to produce more transformative creativity than before. People call it "Augmented Creativity." It is an alternative to the doomsday scenarios of many and a future that I find sexy.

Boris Eldagsen on AI and Creativity

• Home
• Contributors
• About Us

• Data Protection Declaration
• Terms
• Disclaimer
• Privacy Settings

Let's be friends. Follow us:

• 
• 
• 
• 

Further offers from the world of the Goethe-Institut:

• #artbits
• FEHLER
• THE BIG PONDER
• STILLSTAND

© Goethe-Institut 2024