Meet Harshit Agarwal, who uses artificial intelligence to create art
Just as AI’s influencing all aspects of our lives, it is bound to influence the world of art, says Agarwal, a pioneer in the developing genre of AI art
image for illustrative purpose
Using machines, computations, and algorithms in creating art. Hyperbole? Not really, if you are to believe Harshit Agarwal, a pioneer in the developing genre of AI art. Agarwal has worked with AI art since its inception in 2015. His work has been nominated twice for the top tech art prize, the 'Lumen', and he was the only Indian artist at the first global group exhibition of AI art at a contemporary gallery in 2018. Initial examples of AI art were mainly reflected in its ability to create hauntingly familiar yet alien forms. The field in the last five years has deepened considerably and Harshit's work is a testimony to that inquiry. Speaking to Bizz Buzz exclusively, Agarwal shares the fascinating story of evolution of this new art form and his take on how this new forms of art need to be seen in a relationship with human artists
What is AI art? How did this new genre of cyborg artists come into being or how did this new movement called AI art begin?
AI art is a new art movement and new art practice where the human artist works with machine learning artificial intelligence to create art. AI, very simply put, is where a machine learns from a collection of data, the pattern within the data and then tries to reproduce that pattern. As a visual artist working with AI, I work with a lot of images as data that the machine trains on and then is able to produce new images that represents that data set. The process is obviously much more involved in terms of how this artist created and it is in the training process that a lot of subtle variations lie where the craft of the human artist comes into play and also in terms of what data set, they choose to work with and how they collect the data set.
AI art came into being in around 2015 when Google released something called Google DeepDream project, which essentially for the first time exhibited the machine's ability to learn from millions of images, a visual perception of the world is somewhat similar to human so given an image it would be able to identify if there is cat or a car, road, mountain, etc., thousand such kind of categories.
Using machines, computations, and algorithms in creating art may apparently sound like a hyperbole, so how do you make it happen? And what about the acceptability of this art form?
This is a very different kind of art, but not a hyperbole because it's real and I am very much practicing that. It's like training the machine or the process which involves exposing the machine to certain examples such as different kinds of image, text, music, etc., and then asking the machines to learn patterns from that without explicitly giving it any rules of shapes, colours, forms, line. This is purely in pixel to pixel level that learns patterns what pixel goes where from the overall image which is part of the data set not any one image but an overall pattern understanding of all the thousands of images and once it is learned it is able to produce new images that represent the data set. This is where the craft of the AI artist lies to a large extent.
You were the only Indian artist among 7 international AI art pioneers at the first global group exhibition of AI art at a contemporary gallery in 2018 and you have received a number of top tech-art awards globally. What and how has been your global experience of taking this art form forward?
I have been working in the tech-art field for almost a decade now and in the AI art field since its inception around 2015. I also had the opportunity to immerse myself in this new form of art and experiment various media via AI creation of visual, sculptural, text based and over the years I have exhibited work internationally in places like Museum of Tomorrow in Brazil one of the most famous tech-art festival, QUT Museum in Australia and many more places and definitely in India as well. The existence of tech based art and new media art has been more prevalent for quite a few years. In India for me very few artists from the global South working with AI art probably very few are working on this art. After my exposure internationally, I wanted to work for themes that are more central to the reality of my Indian roots. I have been working out of India for almost four years now.
Is it truly the future of contemporary art, as some people say? Can AI replace puritan artists?
I definitely feel this is the future of contemporary art. Primarily because of a few reasons, one is, we are all immersed in technology today, not just because people working in technology or software companies, but each one of us. Like you wake up with your phone, your phone recommending things to you, you shop on Amazon, you drive on Google maps, the news feed you receive, the social media feed you receive, everything is curated by AI for you. At this moment, not everyone knows that they are immersed in AI today. And therefore as an artist, it is very important for me to look at technology beyond its transactional, its efficiency and economic purposes, but to rather look at it emotional, relationship perspective that us as humans need to build with it. Because we are so constantly immersed in it. That is why making art with AI is highly relevant & important today. And I believe just as AI's influencing all aspects of our lives, it is bound to influence the world of art.
Around the world, artists are exploring the interface between art and artificial intelligence, to create works, but are these works being auctioned as well?
These works are definitely being auctioned both Christie's and Sotheby's have done auctions of AI art which have been received very well and successful. I have personally auctioned many of my AI art works as NFT's on a platform called 'SuperRare' which is one of the most premiere curation only NFT platforms in the world. Definitely it's receiving a lot of attention conceptually, financially, economically. But it's a new kind of patronage for people who are excited about the future more than the present.
If MIT starts producing artists, what will happen to the traditional artists with paint brushes?
All kinds of art should be appreciated, maybe traditional art or AI art. It's totally upon the audience how they will appreciate it and I am sure different kinds of art will continue to exist. For me what is exciting about MIT to start producing artists like myself is the fact that it's new and a relevant conversation that needs to be had about technology, role and society. Not just from a productivity, transactional, economic, efficiency perspective but truly from exploring the ideas what the poetics of the machine is and that is what I think my art journey is all about and more sort of these anti-disciplinary artists who work across disciplines to explore a theme are going to become more relevant with the time.
Can these new forms of arts add emotions, expressions, etc? Is this completely AI system-driven or there has to be a collaborative effort between humans and the AI system?
These new forms of art need to be seen in a relationship with human artists. This is very important to point out that the artist in all these works is the human that is myself. The AI is not the artist, there is no intentionality behind the AI. I tell the AI what it needs to train on and I decide how it trains on the data that I set to train on. I define the process exactly and I ultimately choose the outputs from AI that I call art. This whole process is governed by the AI artist and there is an influence of the machine which the artist embraces where the training process leads to output that may not have been purely done what the human artist expected and then they incorporate the influence into the art work or choose not to incorporate that's again the choice of the human artist. So that level of influence is more than sort of your traditional painting on canvas or something. Obviously, all materials influence the artist but here there is a more inscrutable nature of influence because the neural network which is the training process has millions of parameters which humans simply cannot get into the bottom of every bit of it. So that level of abstraction results in a slightly more organic influence by the machine, but the artist remains the human and therefore any emotion, expression that needs to be incorporated into the work comes from the human artist in the choices that they make about data sets, themes, visual aspects of training that they want to work with.