What will prompt Artificial Intelligence art to come of age?

The internet has no dearth of options in AI art generators, be it Artbreeder, Midjourney, NeuralStyler, GanPaint Studio (developed by MIT-IBM Watson AI Lab), DALL-E by Open AI, StyleGan by NVIDIA or the intriguingly named This Person Does Not Exist (creates real-life portraits of people who do not exist). There’s a mobile app, WOMBO Dream, which creates art in different styles. There are more AI generators than what’s listed here and each of them create art with prompts. The importance of giving prompts is such that there are courses on it, making it a significant input calling for human intervention (the foremost human intervention would, of course, be to develop the AI tool).

Art made with AI by Jayanthi Madhukar.

Bengaluru-based new media artist and HCI (human computer interaction) researcher, Harshith Agarwal notes, “AI Art is very much human generated today. It's a human artist that decides how to train an AI, what data to train the AI with and what is the art that comes out from the infinite possibilities that an AI might throw out at you.” He and other artists generating AI art believe this aspect makes AI art very much like other art forms which use other kinds of materials and tools to craft their art.

The Conundrum

If AI art is like any other form of art, can it be displayed alongside the rest of the art done by humans? In 2018, India’s first AI art show, Gradient Descent, was held at New Delhi’s Nature Morte, India’s leading art gallery promoting challenging and experimental forms of art, featuring the AI artworks of seven new media artists. Aparajita Jain, Director at Nature Morte, believes AI art is still working its way into the everyday art creation practice. However, at Gradient Descent, its curators - Raghava KK and Dr Karthik Kalayanraman – didn’t display other art forms. Suresh Jayaram, the multi-hyphenated founder of 1 Shanthiroad, Bengaluru’s edgy and experimental art space, believes it isn’t fair to mix different languages of art. “As a curator, I have no problem in exhibiting AI art but would draw a line at mixing different art languages,” he says. “Eventually, context should make meaning in displaying any art.” Of course, for artists like Agarwal, the context is that AI is just a ‘tool’ and in that capacity, it can be displayed alongside other forms of art making.

A question that arises today is the status of AI art. This brings us to Obvious, a Paris-based collective of artists, Machine Learning researchers and friends interested in AI for Art. One of the collective’s AI artworks, Portrait of Edmond de Belamy, was the first AI generated art to be auctioned by Christie’s New York in 2018. This was a much-needed disruptor in the field of AI art. However, most of the artists, curators and gallerists in India believe that AI art is still at its infancy here. “The consequences of showing AI art in galleries, especially in terms of commercial sales, is yet to be seen in India,” Jayaram says. “If AI art isn’t commercially successful, then the works may not reach the museums which are the gatekeepers of art done by humans.” So far, the practitioners of this genre, according to him, are “people who want to stir up things in terms of visual imagery”.

The other question is the issue of copyrights. Can an artist have copyrights to an AI generated artwork? Since AI has the ability to train on other artists’ visual styles and reproduce them with different content quickly and easily, how can a singular outcome of AI art be copyrighted? The legal landscape is evolving and two schools of thoughts are emerging. One perspective is that AI systems are tools or instruments created and directed by humans. Therefore, the human(s) who create or own the AI system are often considered the authors of the AI-generated art. Advocate Ganesh Shivaswamy, who does copyright litigation, is of the opinion that if original, there can be copyrights on the said art. “All AI is based on prompts. To believe otherwise would mean AI has a distinct creative ability. The argument could be made that human intelligence is also based on environmental and subliminal prompts, which it largely is. So, the prompts for AI are from the human who gave them and therefore, the art could be attributed to that human.”

The other school of thought believes that because the AI tool collects various information derived from the minds of individual(s) and uses methods of permutations and combinations to get to a ‘final result’, it wouldn’t be justified to claim copyrights. On a placatory note, Agarwal says, “More responsibility needs to be taken by artists to avoid misuse of other artists' works, especially in the interim time, till everyone starts more actively knowing about and using AI.”

Famously, the US Copyright Office ruled that the award-winning AI art by Mathew Allen, Theatre D’opera Spatial, could not be copyrighted, reinforcing the fact that the legal landscape has to evolve. “Unless and until someone does something so extraordinary, things will remain fluid in the AI art genre,” Jayaram remarks.

The Future

Art made with AI by Jayanthi Madhukar.

The Allen ruling may certainly prod artists to increase the amount of work they put into art produced by using AI tools. Only with such significant tweaks by a human, which make a significant change to the artwork, can an artist try and get a copyright on the whole work rather than on a part of a painting (as it was decreed in the Allen ruling). This can be a hopeful development boding well for the future of AI art. “I believe AI will be used as a tool more than the finished artwork and that is why I believe the future is very bright,” Jain predicts optimistically. Food for thought: Allen did 624 tweaks or adjustments and was still not allowed to copyright the art.

The bigger question remains: Will AI replace humans in creating art? Jain has yet another answer. “I don’t believe that art gets replaced, it gets added on. Photography added to fine art, it didn’t replace paintings. I believe artwork created using AI will be part of the larger family of art.”

Agarwal, unsurprisingly, predicts an exciting future. “I think each artist, of each kind of artistic medium (not only the visual arts), will use AI in one way or the other in their process. It may not be to create the final piece of artwork, but could be used to mockup something, to iterate quickly on ideas and the hundreds of other aspects involved in a creative journey. I am personally cognizant of AI's deep involvement in all aspects of our everyday lives in ways we don't fully acknowledge or understand. I think it therefore becomes necessary to engage poetically, critically, artistically and holistically with it and look beyond all the hype.”

Instead of gazing into the crystal ball for answers on the future of AI art, the question was directed to OpenAI’s ChatGPT. This is what it said: The future of AI-generated art holds the promise of even greater creativity, realism, and diversity as advancements in machine learning and generative models continue to unfold. Collaborations between artists and AI systems, ethical considerations surrounding authorship and ownership, and the integration of AI into the art creation process are likely to shape the evolving landscape of AI-generated art.

To paraphrase an extraordinary human poet of our times, we believe that this means the answer is blowing in the wind.