“Combinatorial creativity” is a term used to describe the process of generating new and innovative ideas by combining or recombining existing elements in different ways. It involves exploring various combinations, permutations, and arrangements of existing concepts, elements, or techniques to create something novel and valuable. Combinatorial creativity emphasizes the idea that creativity can arise from the reconfiguration or synthesis of existing ideas rather than starting from scratch. It recognizes that the creative process often involves making connections between disparate elements, exploring possibilities through the combination of different elements, and discovering unique solutions through the exploration of various combinations – If I am not mistaken, The Dissertatio de arte combinatoria (“Dissertation on the Art of Combinations” or “On the Combinatorial Art”) is an early work by Gottfried Leibniz published in 1666 to deal with this topic.
The Combinatorial Creativity approach executed through the joint and multiple uses of AI platforms such as Stable Diffusion, Midjourney, Lookx.ai, RunwayML and others, is inevitably thinning that dividing line that separates humans from machines through recognition (or disavowal) of the gift of creativity. However, recent applications of artificial intelligence in the field of aesthetics seem to question this vision, showing how machines can generate original, innovative and pleasant artefacts, imitating or exceeding human performances, humans according to an anthropocentric and romantic vision of the creative act seek to harness and slow down the technological goal achieved. Are we therefore in the century that “threatens” the artistic and aesthetic value of human intellectual and cultural production?
If we look at the history of art, we can see how many styles and innovations are the result of formal and combinatorial rules (let me say also techniques) that could be reproduced or predicted by evolutionary algorithms. Furthermore, we can observe how our aesthetic evaluation of a work of art depends not only on its formal appearance, but also on the historical, conceptual and symbolic context in which it is inserted, and on the intentions we attribute to its author.
I’ll give you my thoughts here, which I have been defending for years even before the clear manifestation of the potential of AI, it is a strong expression probably but it has its foundations, namely:
The work of art does not exist! There are only artistic works. Art is an intimate matter of feeling. The critics create the “work of art”.
Coming to the present day, however…
As artificial intelligence systems grow in sophistication, many researchers are exploring how AI can be applied to creative pursuits like art and design. One of the most promising frontiers is generative AI – using machine learning to create new artistic works autonomously. However, some question whether AI can truly be considered “creative.” Through a technique called combinatorial creativity, emerging generative art systems are demonstrating how AI may extend human imagination in novel and insightful ways.
Traditional views of creativity focus on unique insights or solutions generated through conscious problem-solving. However, researchers have proposed that much of human creativity actually arises through unconscious cognitive processes that combine familiar elements in new permutations. When exposed to the sufficient source material, our brains naturally find unexpected connections and generate fresh ideas through this kind of combinatorial thinking.
Generative AI systems work through an analogous process of combining and recombining visual patterns found in large datasets. By training on millions of human-made photographs, illustrations, and other artworks, these models learn aesthetic conventions and probabilistic relationships between different graphic elements. They can then generate new images by sampling from these learned representations in unpredictable ways.
On the surface, the products of generative AI may seem random or meaningless. But a closer examination reveals how these systems manifest aspects of combinatorial creativity through new mixes and matches of visual motifs. For example, generative models trained on artwork may produce hybrid images blending styles from different eras or genres. They can also visualize concepts that are implied but not explicitly present in the training data, like architectural blueprints that have never been constructed.
These abilities point to how generative AI extends the scope of human imagination in significant ways. With access to far greater volumes of source material than any individual mind, AI systems can find novel patterns and make creative connections at a scale impossible for humans alone. Their outputs reflect the unconscious blending and reassembling of cultural, artistic, and factual influences in a constantly evolving stream.
As generative models continue advancing, their potential for combinatorial creativity will grow exponentially. With guidance from human creators, this emerging form of AI may inspire new styles, spark unconventional ideas, and shape design in architecture, engineering, and beyond. While different from human subjectivity, generative art demonstrates how machine intelligence can augment and expand the frontiers of imagination when guided by combinatorial processes – opening new vistas for creative collaboration between humans and intelligent systems.
Here are some ways that combinatorial creativity in generative AI art could be applied in the architecture, engineering, and construction (AEC) industry:
- Concept Design: Generative models trained on design styles/elements could rapidly generate novel early-stage concepts by combining styles, materials, geometries, etc. This could inspire new ideas early in the design process.
- Form Study: AI could explore unconventional spatial configurations and structural assemblies not easily conceivable by humans to discover new forms and structural solutions.
- Material Exploration: Combining textures, patterns and structural approaches from different materials/structures could lead to new composite or hybrid material discoveries.
- Parametric Design: Generative AI could automatically explore wide design spaces defined by architectural parameters, discovering optimal or unexpected design solutions.
- Detail Design: AI art techniques like style transfer could combine ornamental motifs and construction details from different eras to inspire new embellishment details.
- Visualization: AI could conceptually visualize projects that integrate art, landscape and architecture by combining elements across disciplines.
- Construction Planning: Generative scheduling/planning tools could consider unique activity combinations to discover more efficient construction sequences.
- Heritage Preservation: AI may uncover lost architectural styles by combining elements from existing structures to reconstruct demolished works.
Honestly a real conclusion to this article or rather it is not an article but a personal observation, a topic to put on the table to be discussed at different levels, it doesn’t have a real conclusion, rather it’s more of a beginning and as such I like closing it with a series of questions rather than answers or proselytisms:
How does the concept of creativity change when machines can generate works of art autonomously or in collaboration with humans?
What are the distinctive characteristics of generative aesthetics, i.e. those based on the use of algorithms and mathematical models to create visual forms?
How do works of art generated by AI relate to previous artistic traditions and the cultural and social contexts in which they are produced and enjoyed?
What are the ethical, political and legal implications of the use of AI in art and media?
How can we evaluate the quality and meaning of AI-generated works of art?
Are we running the risk of automating aesthetics and parametrizing the concept of beauty?
Stay tuned! 😉