Track: Undergraduate STEM Research Competition

Abstract

Generative Adversarial Network has been the center of attention in the domain of artificial generative knowledge processes, such as handwriting, painting, and the creative field in general. In this paper, we focus on the human-AI relationship and study how humans perceive and interpret the generative process and outcome of a Generative Adversarial Network that generates handwritten digits. Specifically, we explore the outputs of the handwritten digits generated by the Generative Adversarial Network via NVIDIA DIGITS, as they are perceived by humans. The analysis suggests that humans do perceive the handwritten digits generated by the Generative Adversarial Network to be better over time. Further, the study suggests that human does relate to the handwritten digits generated by a Generative Adversarial Network to a certain extent with around 81.25% of the study participants indicated that the handwritings were written by children who are 9 years and above. We present implications for future interdisciplinary research at the intersection of artificial and human intelligence.