One cannot, in all their seriousness, comprehend what went into writing the “Requiem for a dream” or painting the frescoes on the ceiling of the Sistine Chapel. Art is a consequence; of intelligence and experience. But, what happens when this intelligence is augmented with an external entity, an algorithm?
Artificial intelligence has intruded into the space of creativity, the final frontier of the human intellect, through algorithms such as Generative Adversarial Networks (GANs). GANs have become fertile tools for artistic exploration. Artists such as Refik Anadol, Robbie Barrat, Sofia Crespo, Mario Klingemann, Jason Salavon, Helena Sarin, and Mike Tyka generate fascinating imagery with models learned from natural imagery.
Sign up for your weekly dose of what's up in emerging technology.
Machine learning as a platform to generate art through unprecedented means, has been gaining traction. Whether we like it or not, AI has found its way into painting, architectural designs, making music and more. AI-generated art forms will attract its own tribe. For this to flourish, there is no better place than ML conferences.
The most popular conference Neural Information Processing Systems or NeurIPS, now in its 34th year, has been encouraging machine learning in creativity and design for the past four years through workshops.
And the result is this:
The above artworks are few of the many submitted at NeurIPS conference over the past couple of years. There are demos of music and other works as well. Find them here.
Now let’s take a look at few of the research papers that were presented at NeurIPS:
1| Generative Neural Art By Aaron Hertzmann
In this paper, the author investigates why GANs are such powerful tools for making art. GANs cause visual indeterminacy by creating plausible compositions and textures that nonetheless defy coherent explanation like the ones shown above. The author posits that visual indeterminacy can be understood as a perceptual process, and GANs can act as a potential tool for both arts and for neuroscience experiments.
2| Game Design using Creative AI By Anurag Sarkar
In this paper, the author demonstrates and argues for leveraging creative AI techniques for game design. He trained variational autoencoders (VAEs) on levels from the games Super Mario Bros. and Kid Icarus and showed how the affordances of the learned models could help inform co-creative game and level design.
3| Immersions By Vincent Herrmann
According to the author, Immersions is a system that visualises and sonifies the inner workings of a sound processing neural network in real-time. In this work, the author tries to make the innards of sound processing neural nets audible. This work is all about the generation, visualisation and control of sound in real-time.
This year too, NeurIPS is inviting art submissions in any medium, including but not limited to:
- Dance, Performance, Installation, Physical Object, Food, etc.
In the official announcement of the 4th edition of NeurIPS workshop on art, the organisers stated that the generative models enable new types of images, music, and text, including recent advances such as StyleGAN2, Jukebox and GPT-3.
“This one-day workshop broadly explores issues in the applications of machine learning to creativity and design.”
“The goal of this workshop is to bring together researchers interested in advancing art and music generation to present new work, foster collaborations and build networks with the aim of solving the most pressing problems in the field,” said the organisers.
By having a dedicated segment to creative arts, conferences can nurture a new breed of research-induced art that remains obscure to the current state of our perceptions. Not only this, but the experts can offer their feedback to the AI artists regarding their algorithms while investigating the social and cultural impact of these new models.