While we traditionally think of Artificial Intelligence (AI) as being used as a key component of digital transformation by optimizing operations, creating connected digital products, and enhancing customer experiences, its use cases actually expand well beyond technology. In past posts, we’ve emphasized how AI coupled with people drives innovation. The executives interviewed as a part of the 2018 TIBCO CXO Innovation Survey seem to agree; the human element provides AI with the foundation to be successful. One such example is art, encompassing architecture, painting, and of course, creativity in general.
Image generation using AI technology started circulating heavily in 2015 with the launch of Google’s pattern-finding software, DeepDream. The idea of using AI as a means of artistic creation was relatively young at the time, with little projected sustainability due to the lack of aesthetic and concept to last in the art world.
Recently, auction house Christie’s announced that it would sell artwork that was created with artificial intelligence, turning the art world on its head. The artwork in question is Portrait of Edmond de Belamy (2018), an uncanny, algorithm-created rendering of an aristocratic gentleman. It is estimated to be worth $7,000 to $10,000. The men who were responsible for the creation of the piece used a model called GAN, which stands for “generative adversarial network”. GANs are a recent development in AI, as they represent the next step in the evolution of neural networks: the interconnected layers of processing nodes, modeled loosely on the human brain. This is what makes them so successful and attractive in the creation of art; unlike DeepDream that yield repetitious results based on pre-existing images, GANs can be trained to produce completely new, and dramatically different, images.
For example, The Next Rembrandt is a collaborative project that took a piece from Rembrandt and used AI to recreate it. The team first examined the original painting to get a sense of its surface texture, elemental composition, and pigments used. The next step was to create a height map using two different algorithms that found texture patterns on the canvas surface and layers of paint. The information collected was transformed into height data, allowing the team to mimic Rembrandt’s brushstrokes. To recreate the painting, an elevated printing technique on a 3D printer was used with multiple layers of paint-based UV ink. The final height map determined how much ink was released onto the canvas during each layer of the printing process. In the end, 13 layers of ink, one on top of the other were used to create a painting texture true to Rembrandt’s style.
Another example is the ability to preserve architectural history. Working with “heritage activists” Iconem, a Paris-based company that creates 3D digital models of historic landmarks threatened by war, conflict, time and nature, Microsoft was able to provide Microsoft AI to enable its mission. Advanced algorithms and the computing power of Microsoft AI enable Iconem to quickly stitch thousands of photos into high-resolution 3D models that help experts assess the damage of a landmark. Iconem is using this technique in Syria to access the war damage on its six UNESCO World Heritage sites, including centuries-old temples, mosques, citadels, tombs, and bazaars. The models are so detailed that they can be used to fight archeological looting in Syria by revealing illegal tunnels and clues to missing artifacts destined for the black market. And they are so realistic that viewers feel like they can touch ancient stone walls and walk on storied grounds.
These examples are just the beginning of how AI technology can contribute to the creation of art. When used wisely, it can help people express themselves and solve worldly problems in ways they never thought were possible before. Who knows, AI might give way to an art movement of its own, forever engrained in art history.