How Humanity Triumphs Over AI in the Battle of Creativity
Artificial intelligence has produced a wide variety of art. However, as a top AI researcher notes, these creations will never be truly original. Artists have found a new doorway to explore creativity further as artificial intelligence advances. Millions of realistic images are created every day in a matter of seconds, and these photos are entered into art competitions, where the “creators” of such images even frequently win. However, none of them were produced by humans.
Generative Adversarial Networks (GANs), a development in AI, were introduced five years ago. It used photos to try to produce comparable outcomes. If you gave it a picture of a cat, it would produce entirely different copies that matched. This development was groundbreaking, and lots of artists across the world started to use it, either for their own benefit or as a modicum of inspiration for being creative.
ALSO READ: Exploring the Intricate Behaviors of Robots in Our Evolving World
Then a newer generation came into existence that gave users more control over the generated images by using language to generate them. The model was trained on numerous photographs and the text captions that accompanied them to comprehend how the words related to the images.
The network then notifies the artificial intelligence model if its estimation of the location of the tree and bird in an image, for instance, is right and how entirely accurate it is. The model thus determines which words connect to which photos by doing it for billions of photographs.
Missing Out On The Minute Details
These AI-based models do have some difficulty understanding minute details. Since they are trained to optimize through lost function, a criterion that encourages them to optimize an entire image, aiming to get the bulk of it correct, they will struggle to generate something small in comparison.
ALSO READ: Photoshop’s New Generative AI Feature: A Revolution in Image Editing
The AI overlooks the minute features that humans are wired to notice, such as a hand with four fingers or a figure standing on three legs. This is identical to any other minor background detail in the AI’s eyes.
How Much Energy Does It Take To Create Such Images?
These models need a lot of effort throughout the training process. They require weeks of running time on graphics processing units (GPUs) since they process billions of photos. Then, to streamline the process, you must rewrite them numerous times.
But even once the models have been trained, they must continue to operate continuously on a GPU, which has millions of processors and is an extremely power-hungry technology. They take a substantial amount of energy to run for the required periods, which has negative effects on the environment.
How Do We Regulate This?
How can we manage the information provided to an AI? There are various viewpoints on every topic, including politics, religion, fashion, society, and everything in between. We cannot filter the information provided to support particular viewpoints.
Naturally, AI must encompass all world views and ideologies. There will be a lot of disinformation surrounding this, but that is just the way the world is, similar to how we peruse social media feeds.
Millions of photographs being used online without the artist’s permission raise a copyright issue. The issue is that while it is unethical, it may not be violating copyright laws. Under any copyright legislation, there would be no issue because it is not making direct derivatives of the image but rather transformative variants.
ALSO READ: Google’s Revolutionary Shift in AI Training Policies
The only thing that the present generation of AI can do is replicate human work. To produce something useful, it needs to be substantially under human control. It’s a fantastic tool, yet it lacks creativity on its own.
Conclusion
This appears to be the truth about AI. The guidelines are followed in a creative process that is primarily human, and AI is utilized in the process only to produce content. This movement has been present for the past five years. The works of the early artists had a distinct aesthetic that was strange and unhuman-like.
It had a particular aesthetic, but now everything is turning more photorealistic. It is very good for creating realistic visuals, but it has lost its capacity to surprise, frighten, and produce strange effects.
