When Art Transcends Human Creativity

 

Delhi-based Nature Morte Gallery’s recent exhibition, ‘Gradient Descent’, seeks to establish artificial intelligence as a new genre of contemporary art-making

Ankush Arora

What happens when a statistician and an artist decide to create a new form of art? The outcome is an experimental initiative for artists and technologies of ‘Artificial Intelligence’ (AI) to reimagine creativity in contemporary art practices, while establishing a collaborative human-machine relationship. That is the stated goal of ‘64/1’, the Bengaluru-based art curation and research collective founded by brothers Karthik Kalyanaraman, whose interest in visual arts predates his career as an economist, and Raghava KK, a multidisciplinary artist known for creating an art book for children on the iPad.

And what happens when a gallerist realises that AI (broadly, the ability of machines to replicate human intelligence) is slowly beginning to shape the way we access technology? The result, this time, is a startling collection of seven exhibits and installations, made entirely by AI algorithms (or computer codes), establishing AI as a new medium to make art. This is a rare show curated by an Indian art gallery, joining an emerging movement of AI-based art in other parts of the world.

 

Imaginary Landscape, 2018, Nao Tokui. Courtesy: Nature Morte, Delhi

 

Featuring mixed-media elements like AI-generated sounds, videos and images, the exhibition at Delhi’s Nature Morte Gallery debates the common apprehension that technologies like AI could make human labour or creativity redundant. Instead, human expression in art can be reinvented by technology as a combination of man and machine, the exhibition ‘Gradient Descent’ demonstrates, even though different forms of human labour are being replaced. The exhibition’s six overseas artists and one Bengaluru-based practitioner have trained AI networks to produce the final artworks by exposing them to multiple audio-visual materials.

Here are excerpts from a conversation with Karthik, who taught econometrics in the UK and US for about four years, before setting up his art curation and research collective. He has curated this exhibition with Raghava, who began his career in art as a cartoonist, moved to painting and simultaneously began experimenting with digital processing and programming tools to create art.

How did the two of you get interested in AI art?   

We were both obsessed with art and somewhat frustrated with the lack of deep engagement of artists with what we thought are issues of the future: the rise of AI as a symbol of continuing labour replacement by technology and climate change. So it was natural for us to start looking for artists who were engaged with AI.

What does the title of your show, ‘Gradient Descent’, mean? 

It’s a pun. ‘Gradient Descent’ is the name for a particular mathematical (optimization) routine that the AI algorithm performs in order to learn. At the same time, we want to imply that there is a gradient, a continuum, between man and machine.

Has your show, in any way, changed perceptions towards AI art in India?  

I think most people in the art world did not really realize the developments made by AI art, so it was a bit of a surprise for them. So I’m not sure that it has changed perceptions, because I think initially there was a huge lack of awareness. But I think now we have definitely seen younger artists become more interested in using AI as a tool. As for people in the tech world, our hope is that they have started to realize that their world intersects with the art world in interesting ways and will want to experiment with cross-overs.

Perception Engines, Tom White. Courtesy: Nature Morte, Delhi.

 

How is the AI art market in India shaping up?    

The potential is huge. But the truth is there are very very few artists truly engaged with AI in India, despite the fact that the kind of AI needed to create interesting images is actually taught in computer science departments like say, at Ashoka University. This is because of the wariness with which our art schools still regard this media art space. The challenges are not technical. We have the technology. The challenges are in getting more members of the art world (art school faculty, students, critics, curators) to engage with this space.

Your show is dominated by international artists, with only one Indian artist from Bengaluru. Did you intend to make it a global show or does India not have too many AI artists?    

Both. It was a pioneering show globally too: the first time a mainstream art gallery organized such a show in the world. So we wanted it to be global. As for India, after intense search we only found a few artists in this space and of them only Harshit Agrawal was doing stuff that was both at the forefront of technology and conceptually interesting.

The Anatomy Lesson of Dr. Algorithm, Harshit Agrawal. Courtesy: Nature Morte, Delhi.

 

It appears that while talking about AI-based art, the medium takes the centerstage, somewhat marginalising the actual theme of artworks. And that has also happened in the way your exhibition has been received. Is that problematic or even unfair?

Very much so! I think because of the novelty value of this kind of show, purely the ‘AI’ part is getting stressed. We see this globally too: for instance the piece auctioned off by Christie’s is really quite unoriginal and has been done by other AI artists, so they are primarily selling as made my AI, creating a kind of idiotic mystique around the whole field. The truth is the human artists have gone nowhere! In a group show like ours, you could really see the aesthetic diversity of the artwork and this stems mainly from human decisions, because all the artists are starting out with more or less the same raw algorithm.

The human artist designs the concept, the human artist trains the machine in a specific visual style, the human artist curates the output. So then whole completely produced by AI idea is a myth. So yes, it is unfair.

In India, how are you planning to keep the conversation going regarding AI work?

We are in the initial stages of first moving the show to a couple of other cities. We would love to be able to create an outreach to art schools and engage young artists with the art of the future. Another idea was to create a lab that would provide the resources for young artists to engage with this stuff.

Self-Portrait, Mario Klingemann. Courtesy: Nature Morte

 

According to the exhibition, AI art is just two years old. How did you arrive at this timeline? We do know that artists have experimented with the idea of creating art independent of humans for more than 50 years. 

In a sense you can say (French modernist) Marcel Duchamp kicked off that process with his readymades so I would date it even earlier. But the idea that an image can be made, where one has no control at all over the final look of the image, where the machine is not given any procedure or recipe for creating images, but is only fed a whole lot of examples to learn from, is very new.

The main idea for allowing a machine to generate an image, pixel by pixel, came in 2014 with the publishing of the paper on Generative Adversarial Networks by Ian Goodfellow and co-authors. Before that there was some experimentation with things like style transfer and Google DeepDream, but these were not as aesthetically interesting to artists because of their limited image-making capabilities.

‘Gradient Descent’ concluded at Nature Morte Gallery, Delhi, on September 15, 2018.

---

Ankush began his career as a journalist in 2008, and has since covered multiple stints in print, television and digital media in India. In 2016, he took up a communications and outreach assignment for an American social innovation organisation, which works with the Tata Trusts in India. He is currently working, in Delhi, as a media publicist for art practitioners. He tweets @artandculturediary, and shares his photography on Instagram.