It's a Long Way to Neutrality. An Evaluation of Gendered Artificial Faces

Implementing gender-neutral virtual agents seems to be one possible solution to the problem of designing technologies which do not represent and convey gender stereotypes. Three tests were structured with the intention of selecting faces of male, female, or neutral gender hypothetical virtual agents. In each of these tests 30 participants assessed the gender and age of 9 hypothetical virtual agent faces by means of an online questionnaire. From the results of these tests, 3 faces were selected, one male, one female and one neutral, which were assessed through an online questionnaire (N = 83) with reference to some feminine or masculine characteristics of their personality. The willingness/pleasure to interact with artificial agents having those faces was also assessed. The results highlighted the difficulty in synthesizing faces that are perceived as absolutely neutral. Evaluations of the stimulus characterized by greater gender neutrality were less likely to refer to a female stereotype. The stimulus representing a gender-neutral face resulted also less accepted and liked than the male stimulus in all aspects considered and, in fewer aspects, than the female stimulus as well.