When it comes to AI, everyone has a crystal ball. So do I. Today, I asked mine to tell me what the future holds for all visual imagery being created by the likes of Midjourney, Stable Diffusion, and their cousins in the world of AI.
It thought long and hard—and this is what it said:
Generative AI will emerge as an independent art form, like painting, photography, digital painting, and visual effects. It will draw inspiration from these existing disciplines but mimic none of them.
Knowing well my penchant for a reasoned answer rather than mere soothsaying, it went on:
Let’s examine some traditional and current visual creation mediums along multiple dimensions and see where AI art fits on them.
Abstraction vs Realism Axis: A Picasso cubist painting is his imagination on canvas, with only a passing reference to the human form it depicts. Contrast this with Steve McCurry’s photograph of the Afghan Girl or Jeff Widener’s photo of the Chinese man standing during the 1989 Tiananmen Square uprising—they are as real as it gets. A painter lives in the abstract while a photographer creates very close to the physical world. Perhaps a VFX artist operates somewhere in between, often moving back and forth along this axis. What about AI art? While future advancements may blur these distinctions, at current levels of technology, AI art is easily distinguishable from a photograph. But this does not reduce its aesthetic appeal. As time goes by, AI artists will likely embrace the unique characteristics of their medium rather than be embarrassed by them, positioning their creations at points along this axis away from the extremes of total abstraction and physical realism.
Degree of Control and Flexibility in the Creative Process: A painter has maximum control, manipulating every brushstroke and colour as desired. A digital artist has a vast, yet finite, array of controls embedded in software menus. A photographer works with the few controls offered by the camera. When it comes to Generative AI, the artists have the least control. Even with rigorous prompt management, they must accept how the model interprets their textual or voice commands.
Human Instruments of Creation: Painters convey their ideas using their hands. Photographers use their eyes, looking through the camera lens. AI artists only need to type or talk. These different human mechanisms used in the creative process suggest that the art forms will also be inherently different, each with its own evolutionary cycle. The simplicity of AI art creation may invite a broader range of people to explore and contribute to this new medium.
Iteration and Reproducibility: Most painters create iteratively, starting with basic sketches and refining them until they are satisfied. They can redo parts of their creation if needed. However, even an accomplished painter would find it difficult to reproduce their own work. A photographer captures images instantly and can quickly decide if a retake is necessary, though their ability to make selective corrections is limited. AI artists, akin to painters, require iterations, but like photographers, their output is available for immediate evaluation. Their ability to correct parts of the creation falls between that of a painter and a photographer. Additionally, AI artists can never be certain that reproducing their work will yield identical results.
At this point, I wanted to get into the topic of copyrights but the crystal ball ran out of battery life. I will ask it more after I have recharged it.
But you get the drift. We have a new art form being created as we speak albeit with some ifs and buts.
No comments:
Post a Comment