Over the past year, we have seen some wild concoctions online, including Joe Biden breakdancing and Spider-Man rendered as a 15th-century oil painting. Maybe you’ve used a filter on Instagram to transform yourself into an anime character. These play trends have recently prompted fierce backlash, clogging our feeds with ‘Anti-AI Art’ posts. So, what is fuelling this response– and why does it matter?
How does AI art work, anyway?
When a user feeds a prompt to an AI Art generator like DALL-E or Midjourney, the machine learning model scans billions of images across the internet and assembles images from patterns between images and their text descriptions. Note that AI doesn’t see images like we do, but as numerical data. Each image is broken into pixels, represented by numbers that represent colors, shapes, lines, etc. AI Art has been welcomed for its ability to generate new ideas, streamline creative processes, and eventually democratize art.
The algorithm and the archive
Behind the stylized outputs generated by software like DALL-E, Midjourney, and Stable Diffusion lies a massive ethical blind spot. They source their images from the internet without permission or attribution to the artists. These artists have now claimed to see their signature style being replicated in AI images, as well as remnants of actual artist signatures. Founder and CEO of Midjourney, David Holtz, even admitted to not seeking consent for sample images being used, stating, ‘There isn’t a way to get 100 million images and know where they’re coming from.’ Not only is this a way for artists to see that their work is being used without permission, but the site also brings up other ethical concerns. Lisa Hakila wrote that when she looked up her ethnicity, all she saw was pornographic content. Because the images being used to train AI come from the internet, they are filled with the internet’s predisposition to pornography, sexism, and racism. This has led to AI art, specifically that of humans, to be very sexualized, particularly women, as well as very whitewashed. There are serious ethical concerns when you think about the ease at which deepfakes or revenge porn can now be created.
On top of that, a prompt for a nurse and assistant shows women, while a prompt for a boss shows white men on OpenAI and Google Research, affirming their gender and racial biases. Even when searching for royal raccoons in OpenAI, the result is that of raccoons in western-style royal outfits without any descriptive reference in the prompt other than royal. AI art based on faces is anglicizing individuals of color, thinning individuals’ faces and over-sexualizing women.
However, the founder of Stability AI argued that this isn’t a problem for AI to fix because AI models don’t create content from scratch. It is a consumer of the internet, and if it is defaulting to whiteness and oversexualizing women, it’s not an error in the algorithm – it is just what the internet has overwhelmingly shown it.
Authorship, theft, and creative labor
AI art isn’t created in a vacuum. These models are trained on millions of artworks pulled without consent from artists who have spent years honing their craft. It’s not up for debate—it’s a land grab in the digital age. Sarah Anderson, Kelly McKernan, and Karla Ortiz are some artists who have filed lawsuits against AI image firms. They sued Stability AI, DeviantArt, and Midjourney for Stable Diffusion use and what Matthew Butterick refers to as ‘a 21st-century collage tool that remixes the copyrighted works of millions of artists whose work was used as training data.’
In a paper published in 2019, McCormack said that with AI Art, the artist becomes a meta-creator. The human role has shifted from directly creating the art to directing another tool that creates it. This raises the question: Is the artist the one who creates art, or the one who creates the system that creates art? He further explains that agency over AI art is diluted since it is generated through algorithmic processes that integrate art already available on the internet, challenging the very idea of individual authorship.
While most of us perceive AI as a powerful tool that lowers barriers to art, Marcus and Davis argued that we overestimate its abilities. Current AI does not think or feel, it cannot imitate human connection. It is statistical, not intentional. It does not create; it just remixes patterns found in data. And while we are forced to rethink what authorship means, current AI is not centered around human values, and the hype around it can be misleading.
Rise of deepfake and erosion of consent
What began as entertainment has quickly mutated into an assault on privacy, consent, and identity. Deepfakes leverage the same AI models to create nearly indistinguishable videos or images of individuals doing or saying things they never did. From revenge porn to political engineering, the consequences are no longer hypothetical. AI software has never been easier to use to manipulate someone’s face, body, or voice, and since the tools are available and unregulated, anyone can use them. For women, marginalized communities, and public figures, this means that online harassment is possible on a scale never seen before.
The Ghibli backlash
OpenAI unlocked image generation abilities, leading to a plethora of new uses. The most popular one is the transformation of real-life pictures to animated caricatures in Studio Ghibli’s art style. Studio Ghibli is a Japanese animation company founded by Hayao Miyazaki. This company is celebrated across the world for high-quality animation and filmmaking. While one camp of internet users went wild with this new trend, the other was infuriated. This camp kept referring to a remark made by Miyazaki in 2016, where he claimed that AI art is an ‘insult to life itself.’
The first camp felt that a ‘style’ can’t be copyrighted, that people should be allowed to have fun, and that if anything, art was being made more accessible to everyone. The latter, especially fans of Studio Ghibli, argued that art is already accessible to everyone. Art isn’t an opportunity; it is a skill, and if it is a skill, then it can be learned. Ghibli films showcase labour-intensive, hand-drawn animation, a craft these artists have spent years honing. It took animators at Studio Ghibli about a year and a half to animate a four-second crowd scene in the movie ‘The Wind Rises’, released in 2013.
Conclusion
Our traditional understanding of artwork is tied to legal, economic, and moral structures, and AI is challenging all of them. When a machine creates, who owns the art? Who gets paid? And who takes accountability for the damage? Some argue that AI is a tool that makes art more accessible, like a paintbrush, while others say that it is an oversimplification of the cultural and legal implications it imposes. AI art has forced us to redefine what authorship even means and asks us to explore more collaborative methods of creating art that integrate human and machine. Humans must steer the process of creating transparent and ethically guided AI systems, acknowledging the devaluation of creative labour at hand.
Written By – Anvesha Sachan
Edited By – Neelambika Kumari Devi
#Theeconomictranscript #TET #Mithibai #AI #Art #Society&Culture
