Was Leonardo Da Vinci the Precursor to Modern AI Image Generation?
What if I told you that Leonardo Da Vinci, the mastermind behind the Mona Lisa and The Last Supper, was centuries ahead of his time, not just in art, but possibly in an area analogous to modern artificial intelligence techniques? Could Da Vinci’s techniques of finding inspiration in random patterns mirror the process of today’s AI image generation models?
Da Vinci’s Stained Walls and Clouds: The Art of Seeing
Da Vinci had a unique technique for stimulating creativity. He would gaze at stained walls, listen to the random notes of the bells, and look at scattered clouds, letting his mind wander and form patterns. These seemingly random stimuli served as an imaginative springboard, leading him to create some of the world’s most admired and studied pieces of art.
This practice, akin to pareidolia, involves perceiving a specific, significant image in a random or ambiguous visual pattern. Da Vinci’s detailed notebooks suggest that he took inspiration from these perceived images, using them as the basis for his artwork. The pareidolia technique allowed Da Vinci to ‘train’ his mind to observe, create, and innovate, very similar to how an AI model might learn from a training dataset.
Image Generation in Artificial Intelligence
In the world of AI, there are models specifically designed to generate new data samples that resemble their training data. These generative models, which include variants like Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs), can produce remarkably realistic images, text, and even music.
A diffusion model (like Stable Diffusion), is a type of generative model that produces new samples by starting from a random point in data space and gradually refining it until it becomes a believable sample. This ‘refining’ process might be seen as the model “seeing” and interpreting patterns in the ‘noise’, not entirely dissimilar from Da Vinci’s pareidolia technique.
Bridging Da Vinci and AI
Now, was Da Vinci using a form of “Stable Diffusion Image Generation” when he stared at stained walls or clouds? Not exactly, given the specific mathematical and statistical underpinnings of AI models. But there’s a certain conceptual parallel that’s worth exploring.
Da Vinci was training his mind to see order in the chaos, to imagine figures and scenes from the random stains on a wall, just as an AI model trains to generate realistic images from random points in data space. Da Vinci took detailed notes, refining his initial observations into complete works of art, mirroring how a diffusion model iteratively refines its initial random point into a believable sample.
Conclusion
The imaginative techniques of Leonardo Da Vinci and the computations of AI models might be separated by centuries and disciplines, yet they share a fascinating similarity: they both navigate through randomness to find, or create, something meaningful. It’s a reminder of how human creativity and machine learning are not so different after all. They both require a spark — whether from a stained wall or a point in data space — to ignite the process of creation.
References
- Suh, H. A. (2013). Leonardo’s Notebooks. Black Dog & Leventhal, New York, NY.
- Al Jalammar, J. (2023). Illustrated: Stable Diffusion. Retrieved June 29, 2023, from https://jalammar.github.io/illustrated-stable-diffusion/
Do you identify as Latinx and are working in artificial intelligence or know someone who is Latinx and is working in artificial intelligence?
- Get listed on our directory and become a member of our member’s forum: https://forum.latinxinai.org/
- Become a writer for the LatinX in AI Publication by emailing us at publication@latinxinai.org
- Learn more on our website: http://www.latinxinai.org/
Don’t forget to hit the 👏 below to help support our community — it means a lot!
