Can AI-generated art be free from human biases?

Around the end of last year, social media spaces were trending with Lensa-generated images of online users. A subscription app, Lensa, makes graphic portraits, called “Magic Avatar” images, using selfies uploaded by its users. Celebrities worldwide stepped in to show how they looked so perfect in their avatars in a “Lensa world”. However, a few days later, hundreds of female netizens worldwide started flagging issues with their avatars. They pointed out how their avatar images had their waists snatched and showed sultry poses. Even after these women uploaded different pictures, Lensa generated hyper-sexualised, semi-pornographic images. As AI takes a strong Foothold over the realm of art, are we equipped with mechanisms to define what is right and what is wrong in this domain in the first place?

AI art is any art form generated using Artificial Intelligence. It uses algorithms that learn a specific aesthetic based on textual prompts and, after that, go through vast amounts of data in the form of available images as the first step. In the next step, the algorithm tries to generate new images that tally with the kind of aesthetics that it has learned. Here, the artist becomes more like a curator who inputs the right prompt to develop an aesthetically-fulfilling output. While artists use brush strokes in other digital platforms like Adobe Photoshop, in programs like Dall-E and Midjourney, all it takes are keystrokes. Say, for example, the generation of an artwork like “Starry Night” in the digital era. While Van Gogh would have taken days of effort to conceptualise and get the correct strokes and paint, in the AI ​​art era, it is just a matter of the “right textual prompts”.

The question of whether AI art is causing “a death of artistry” was raised, last year, when an entry called “Théâtre D’opéra Spatial” generated from Midjourney (an artificial Intelligence program) by Jason M Allen won the Blue Ribbon at the Colorado State Fair. AI artists like Allen think finding suitable prompts to create an artwork amounts to creativity and qualifies AI art as genuine or authentic. In June 2022, Cosmopolitan Magazine released a special report with the world’s first AI-generated cover page, a Collaboration with Dall-E2 and digital artist Karen X Cheng. The Academy Award-winning director of Shape of Water, Guillermo del Toro, however, thinks that AI art is an insult to life. On a question about AI art by the media, he responded, “Art is an expression of the soul. At its best, it is encompassing everything you are. Therefore, I consume, and love, art made by humans.”

As AI-generated artworks derive from a database of pre-existing artworks, there are possibilities of copyright infringement. AI-generated artworks have the potential to benefit from open artworks to generate new AI art that may be copyrighted in the long run. Also, AI artworks may make contemporary artists more vulnerable in a pre-existing framework of underpaid artists. Will it make art a mass-produced commodity that is cheaply available in the market?

Art is one of the few pursuits that makes life meaningful. It remains to be seen if AI-generated art will alienate the experience of art from the masses. AI-generated art dehumanises artworks. Perhaps the most satisfying aspect of generating an artwork lies in making it. It is also doubtful whether AI art will capture the most subtle of human emotions. How much humor is “humorous” for AI? Can AI express grief and pain in the most profound ways as described by our poets? Can AI Capture the enigmatic smile of Mona Lisa that makes one believe that she is shrouded in mystery?

As a technology, AI works on the promise that the more data we provide to AI platforms, the better they perform. There is bias in this data available for AI inputs due to a lack of representation of the less privileged communities — women, people of color and other marginalized groups. Most of the training data for AI art currently emerges in the Global North and is often mired by the stereotypes of ableism, racism and sexism. Historically, art has performed a political function as a venue for dissent. Can AI art overcome these inherent biases in data to bring out Meaningful political engagement?

Some artists believe AI art could democratize the art world by removing gatekeepers. They believe AI will open doors to new art forms and endless possibilities. In combination with technology like 3D printing, AI can generate innumerable derivatives that may qualify as “sculptures” and “art installations” of the future. History says there was a chaotic relationship between art and technology since the beginning. Fine art, however, has persisted through generations despite threats from digital art. What needs to be seen is whether AI art can evolve out of its prejudices.

The Writer is an IAS Officer of the Kerala Cadre, currently posted as District Development Commissioner, Kozhikode


Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button