Artificial Imagination: Is it just an illusion? (2/2)

Artificial Imagination

As present time felt obsolete compared to the new realm of AGI everywhere, I started wondering about the possible consequences of Artificial Imagination and Intelligence on human creativity and subsequently on the arts in the post-Anthropocene. From the technical realm of Samuel Doogan’s computing scope throughout the critical positioning of Johannes Bruder’s research, these discussions inquire topics such as whether AI could create art? What could be the philosophical, political and sociological "consequences" of such progress? And Could this change our anthropomorphic vision of creativity? Hoping you will enjoy the speculative ride

I met Johannes Bruder at the critical media Lab/IXDM (Academy of art and design FHNW) where he researches how infrastructures and technologies support epistemologies & empiricisms in art, design, science and their (sub)cultural distortions.

Juliette Pépin: Do you think it is possible to create an ontology of art which could be used as a training framework for an AI?

Johannes Bruder: I don’t like to think art apart from social context therefore it would be hard to effectively train an AI on an ontology of art, if something like that even exists. At least in the cultures and societies of the Global North, we define ourselves as artists when we believe that we are making art, if it is our profession and if others are recognizing it as such. We can have endless discussions as to what ‘real’ art is, yet the perspectives and standpoints that subtend this discussion will be ever so different and incompatible. In fact, we must study art to recognize and make it; we start with art education in school, then we gradually learn to understand how the art market and its institutions has made a fair point defining what we perceive as art. What we currently understand as AI - particularly deep learning systems - is quite good at figuring out such contextual factors and might thus, after analyzing a corpus of data wrought by contemporary art, be able to come up with ideas for new art projects that would probably not be distinguishable from those of ‘real’ artists. (e.g. our common failure friends’ Predictive Art Bot project.

Tweets from the Predictive Art Bot project: by Disnovation 

JP: If AGI is reached, do you think that an AI could be creative and have imagination?

JB: I find it much more interesting, as you suggest, to talk about the concepts of creativity and imagination in this regard. I would rather not delve too deeply into the discussions around ‘real’ imagination that emerge as Google DeepMind’s deep reinforcement learning systems have successfully taught themselves to play Go without any human input. Since then, the public discourse has been divided over the question whether this means that algorithms are imaginative in an anthropological sense or whether they are just blindly executing programs. I believe that both camps are right and wrong at the same time. Algorithms are indeed characterized by their restlessness and hunger for optimization. As Demis Hassabis, CEO of Google DeepMind told a reporter of the Guardian, his algorithms “don’t even have Christmas off”. We humans, however, seem to be not so different. For instance, to become a professional Go player, one needs to exhibit an exceptional ability to fail without surrendering, as in the process of creating art: we develop techniques and learn to master technologies, we stay up all night and hope for the cathartic incursion, we intoxicate ourselves to get a better view on what we are about to do. Essentially, then, we produce art by constructing and employing systems that force us to leave the beaten path and produce the illusion that we are looking at the world ‘from beyond’.

Google's AI AlphaGo Is Beating Humanity At Its Own Games (HBO)

You might want to subordinate these techniques and technologies to ‘artistic infrastructure’ yet I’d like to add an artist’s perspective from the genre of ‘Intelligent Dance Music’ to suggest a different understanding. In an interview with the online music magazine, the probably most ‘artistic’ electronic music act alive, Manchester duo Autechre, compare their software setup to an artificial intelligence adding that it’s “about as much like an entity as a shit AI in a game is.” What is more important to the artists than the technology’s capacity to match our human understanding of intelligence and creativity, however, is that of the system capabilities to dictate what the music is like and challenge the beaten paths of the artistic mind while preserving the idea of an artwork in a modified form. I feel that this idea of artistic infrastructure creates an interesting understanding of how technology and formats have always disciplined the production of art and the artist. Have we ever produced art that was not technological?

Feed1 (elseq 1) by Autechre

JP: If art is the expression of a form of “self”, is it likely that an AI develops subjectivity based on neural networks? Or would such art be the expression of a new collective objective-ness built upon the data used for training AI?

JB: I don’t think art should be perceived as subjective, since it is scaffolded by the labor, the ideas, and the knowledge of so many Others, whether human or machine. Thus, art is the result of a collective process that is being provided by and channeled through an artistic personality (by way of all the power games constitutive of the art world we all know too well). Without wanting to resort to the somewhat tired ANTtake on non-conscious collective action, I’d like to stress that technology - whether a simple paint brush or a deep learning algorithm - exhibits and alters our subjective cognitive bias and makes us think and act differently. 

JP: If AI were to be legitimately recognized as creators, do you think it could lead to a post-anthropocentric type of art?

JB: Contemporary AI will further change how we think of ourselves as artists and it might be better suited to challenge our anthropocentric understanding of the production of art than earlier technologies. But the decisive aspect might not be that it is as imaginative, creative, artistic as we ourselves believe to be, but that it shows us more clearly how we never had full control over the works of art we claim to have produced single-handedly. It might show us that art has always been post-anthropocentric in that it productively estranges us from being all-too-human. 

JP: This is a rather personal question, yet I am curious to know if you are rather skeptical or enthusiastic about AI making art?  

JB: I am not enthusiastic about the possibilities of AI at all, this is to a lesser extent due to the very technology rather than to the ways it is currently framed. We are now in the midst of a process where our world is rearranged around machine learning systems (e.g., in smart cities) and experimental forests, which is, I think, a next step in a development that began halfway through the 20th century. Since then, we have been augmenting habitats with sensors to create data but now we are designing sensor worlds that need to be augmented with humans. This rather dystopian present will of course provoke some interesting developments as well and the unsettling of our self-understanding as humans, researchers, artists, designers, etc. is one of them.

“Where the city can’t see” An example of a Smart City Dystopian scenario by architect Liam Young at Tomorrow Thoughts Today
Read: Liam Young - "I don't think there's anything new or radical about VR"

Juliette Pépin