Conducting Legitimate Research Using Various Synthetic Imagery From Artmaking Generative AIs

Abstract

For researchers, in-world phenomena offer many opportunities to learn in systematic ways (through various types of observation, research, and analysis). One phenomenon that can bear higher levels of insight involves artmaking generative AIs, not only in terms of how the systems work and are designed, but in terms of their output images. This work asserts that AI-generated imagery may be informative of the underlying training imageset, human culture, design, and symbolism on one level, but beyond this, offer insights about in-world phenomena. This work suggests that as artmaking generative AIs advance (and some of the more sophisticated ones now), the output imagery and imagesets may be interpreted more deeply for insights about not synthetic versions of the world but of the world itself. Precise proposals are included in this work. Both manual and computational analytics methods are proposed. And there is a proposed approach for validating/invalidating the perceived insights from the imagery.