Note to readers: Hello world is a program developers run to check if a newly installed programming language is working alright. Startups and tech companies are continuously launching new software to run the real world. This column will attempt to be the “Hello World" for the real world.
A fortnight from now, art enthusiasts and connoisseurs in the city of Kolkata will witness a new type of art and artist. At the Emami Art gallery, Harshit Aggarwal, 29, will exhibit his work titled ‘EXO-Stential – AI Musings on the Posthuman.’
At first look, you’d think the vivid imagery is solely the artist’s imagination. But here’s the twist: this series of artwork is not just Harshit’s work. He’s co-imagined it with an artificial intelligence.
“I’m trying to engage with our inevitable techno-centric reality and not just watch it from the sidelines. I’m exploring themes such as biases, both individual and societal, gender and agency with AI,” Harshit, who is gearing up for the exhibition, told me last week.
The exhibition, put together by art curator Myna Mukherjee and backed by Engendered, an arts and human rights organization, in collaboration with 64/1, an art collective founded by brothers Karthik Kalyanaraman and Raghava K.K., will be India’s first solo exhibition of AI art.
Aggarwal has been working at the intersection of art and artificial intelligence for nearly six years now, making him one of the earliest to do so. I spoke to him first in 2019, while researching a story on a handful of India’s cyborg artists. At the time, he was one of the six AI artists from around the world to exhibit at a gallery in New Delhi.
The mixing of art and artificial intelligence started sometime in 2014, when tech aficionados started using Google’s DeepDream or creating their own algorithms to create art. Early on, it was about running a large data set of a particular style of art through a neural network and then applying that learning to create new works of art. This process, more widely applied, is also called transfer learning.
But then, in 2014, scientist Ian Goodfellow introduced a method called Generative Adversarial Networks or GANs. Here, a combination of neural networks tries to learn and create completely new works of art. AI artists decide on the training data, a set of algorithms and pick out a set of images as per their aesthetic choices.
Often, these works can also be interactive. For example, at the upcoming exhibition, Harshit will have on display one of his early pieces where an AI algorithm is trained to recognize objects from around 1 lakh images from an open source database (ImageNet). When a human starts to draw something, the artificial intelligence identifies it as the objects it knows and finishes the drawings. Another algorithm then alters stylistic elements in the final drawing.
The idea here is to let humans experience the “continuities and discontinuities” between the human and machine. “Our visual categories are filtered and estranged through the associations the machine actually makes. We are left to wonder whose was that final drawing,” says Harshit.
While artificial intelligence has made a lot of progress in the last 10 years, and we have it in our midst in the form of smart assistants and almost self driving cars, many people will experience it more deeply through art and other forms of expression.