One artist's personal reflections on methods and ethics of creating mixed media artificial intelligence art

A model output on the left contains outputs in the full color range, while the model on the right contains only blue-green hued images, due to the filtering step shown in the middle which uses k-means pixel clustering and HSV filtration to subset training data.
Using FFMPEG to extract video frames from a collection of animated movies to train a model. After the first iteration (left), the training data is filtered down using k-means pixel clustering into 5 groups, then filtering down to images with pixels in an HSV range within the mid to light blues and greens, to train a new model (right).
I make a scientific contribution of my subjective experience as a single unit of self-described 'artist' leveraging artificial intelligence as an assistive visual creation tool, in the hopes that it may provide some inspiration or deeper meaning for fellow artists and computer scientists in this medium. First, I provide some background on my personal history thus far as an artist. Neither artist nor scientist can exist in a vacuum, so I then provide some related work that has helped me contextualize my own work and thinking in this area. I often consider my methods in the creative process chronologically, so I have divided that section according to the loose structure of my artistic workflow. These foundations provide a fertile grounding for discussion around topics of subject matter, reception, community, and ethics. I conclude with some ideas for future work in the realms of theory of authorship, explain- ability tooling, and research framing.
PDF | Preprint | DOI | BibTeX

Data Visualization @ Khoury — Northeastern University
West Village H, Room 302
440 Huntington Ave, Boston, MA 02115, USA