What exactly is creativity?
To my subjective opinion - aren’t all creative work constructed by some form of inspiration? Isn't the nature of creativity an act of imitation? - a synthesis between two ideas to generate a novel idea? To create means to bring new ideas or things into existence.
I did exactly that but with art and photographs. Can you guess the art, style, and artist my 'creative' AI ‘imitate’ to create these ‘original’ two artworks?
This AI learns Post-Impressionism by imitating Vincent van Gogh’s The Starry Night and Paul Cézann’s Mont Sainte-Victoire.
It uses neural representations to separate and recombine content and style of arbitrary images, providing a neural algorithm for the creation of artistic images. ‘A Neural Algorithm of Artistic Style’ by Gatys et al. (1) in which their research inspires “a path forward to an algorithmic understanding of how humans create and perceive artistic imagery.”
Art is inherently logical, you might think that you are being subjective, but your subjectivity is never fully detached from your material conditions. The medium of a work of art creates constraints on what can be portrayed, by instance, an artist may only have certain pigments available, and while they can mix them to generate new colors, the amount of colors they can produce is limited.
The philosophical distinction between logic and creativity (in the sense you mean) is very new, it only came about in the past half-century or so.
Historically, there are three branches/types of logic: Deductive, Inductive, and Abductive. It's Abductive logic (sometimes referred to as "inference to the best explanation" or "hypothesis formation") that we’re referring too, when we talk about that kind of creativity.
The limits aren't even restricted to the medium. The things you have access to dictate what you can possibly portray in your mind. You can't think of a color that you have never seen by instance. Furthermore, your experiences have an impact in how you feel and think. Culture also has an impact.
[1] We compute the difference between the content loss of the photograph and the artistic style image by the Mean Squared-Error loss function. (VGGNet has built in higher-level features to allow such content loss to be computed)
[2] Then, we calculate the derivative form of the squared-error loss function with respect to the activations in the layers. (for backpropogation algorithm)
[3] The computations of the style losses is dependent on the Gram Matrix. Such technique measures the similarity in style between two images.
[4] Now, calculate the style loss for each layer.
[5] Then, calculate the loss of all layers (style loss) for both the original and generated image.