Humans, artificially intelligent agents, and objects are not evaluated differently
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
How are humans, objects, and artificially intelligent (A.I.) agents evaluated differently, and how might mentalizing ability modulate that evaluation? Humans make evaluations based on group membership and similarity, favoring those who are more like themselves. Therefore, it is hypothesized that positive bias will increase with agency, with humans being evaluated relatively better than A.I., which will, in turn, be evaluated relatively better than un-intelligent objects. It is also hypothesized that individuals with less ability to infer other humans' emotional state or less mentalizing ability will not show as big of a difference in bias when evaluating artificial intelligence and humans as compared to individuals with higher mentalizing ability. Participants (n=140) on SONA completed an evaluative priming task, A Reading the Mind in the Eyes task, and questionnaires. No significant main effects of priming or modulation of mentalizing ability were found. This study was limited in the stimuli used for the A.I. representation and in the inherently neutral valence of all stimuli. Exploring whether humans and A.I. are evaluated differently is worth further investigation. ☐ Keywords: evaluative priming, artificial intelligence, mentalizing
Description
Keywords
Evaluative priming, Mentalizing, Artificially intelligent, Individuals, Stimuli