The Artificial Muse - an Inspirational AI
The Artificial Muse is a cloud-based modular machine learning system combining different services to e.g. extract and learn features such as colors, compositions, contrasts, structures et al. from Lipski's pictures.
Fed 3 years ago on an initial set of digital images the Muse constantly gets retrained on new pictures, intermediate results and the information if and how newly generated pictures inspired Lipski in his daily work.
The Muse learns what inspires the painter.
Like a kaleidoscope the Muse generates a multitude of new motifs, ideas and inspirations. A kind of mirror cabinet in which Lipski's current, daily and past work gets reflected, broken and reassembled. In other words - Lipski inspires himself again and again with the help of his Muse, from his own work and creation, in a way he has never seen himself before.
Lipski learns what inspires him.
At certain times the endless stream of muse images can be explored via www.aironair.de.
The Artificial Muse is unique in the world - an Inspirational AI. Contrary to what is called Creative AI, aiming to develop AI systems that are capable of creating "works of art" - photos, images, texts, scripts, music..., Inspirational AI helps to become a new artist or (art) worker. The Muse pushes Lipski, challenges him, becomes part of himself, in dialogue, in symbiosis, at the "in-betweens" and every connection.
We are currently further developing the Muse, taking it to a new conceptual and technical level. Starting point are the developments in the field of GANs (Generative Adversarial Networks), which enable the targeted generation of "data" (and thus things) through the coupling of several neural networks.
The idea is to develop a new tool, an "instrument", like a new generation of brush, seamlessly integrated into Lipski's studio environment, his daily work. The current prototype consists of different data sets and models (Lipski's current works + his old works + the innumerable images of the muse of recent years + specifically selected pictorial material to learn forms, structures, colors) as well as the option of using video input as material. All these components - similar to a modular system in music production - are interconnectable to form (feedback) loops and pipelines.
In the future, it will be a question of combining human and artificial intelligence in their respective strengths, researching and designing the connections and spaces between them. And as is often the case, art is a pioneer.
Lipski is one of the first in the world who collaborates daily with his/an AI, to be creative together, to explore a new new. That for nearly 3 years. We have already learnt a lot and are learning new things every day. Lately, that Lipski and Muse, man and machine, begin to develop their very own language, word for word, brush stroke for brush stroke, and that even exponential times sometimes need time.
Thanks to Open Source & Community
As all good things in the field of AI our muse is based on open source technology, libraries and frameworks such as Python, Pandas, Tensor Flow or Torch incorporating concepts such as StyleTransfer, Pix2Pix or GANs.