摘要
人们对AI的道德判断不同于人类,这可能源于AI的算法身份和其引起的社会认知。道德判断的CNI模型能够同时关注到两者。本研究基于该模型,通过三个实验,探究了身份(人类/AI)、AI拟人化(高/低)对道德困境判断的影响,以及感知温情在AI拟人化与道德困境判断间的中介作用。结果发现:(1)相较于人类,人们对AI的道德选择的结果收益更敏感,对道德规范更不敏感;(2)AI拟人化通过提高感知温情增强人们对AI遵守道德规范的预期。结果表明,拟人化能够减弱人们对AI算法身份的关注,避免人们基于工具理性判断AI的道德选择。本研究对促进AI向人类社会融入具有积极意义。
People’s moral judgments of AI differ from those of humans,which may stem from AI’s algorithmic identity and the social cognition it elicits.The CNI model of moral judgment allows simultaneous consideration of both factors.Based on this model,the present study conducted three experiments to investigate the effects of identity(human vs.AI)and AI anthropomorphism(high vs.low)on moral dilemma judgments,as well as the mediating role of perceived warmth between anthropomorphism and moral judgment.The results revealed that:1)compared to humans,people were more sensitive to outcome-based utilities and less sensitive to moral norms when evaluating AI’s moral choices;2)AI anthropomorphism enhanced expectations for AI to adhere to moral norms by increasing perceived warmth.The results suggest that anthropomorphism can reduce people’s focus on AI’s algorithmic identity and prevent them from making moral judgments about AI based on instrumental rationality.This study holds positive implications for promoting the integration of AI into human society.
作者
常扩
周肃军
周详
CHANG Kuo;ZHOU Sujun;ZHOU Xiang(Academy for Advanced Interdisciplinary Studies and School of Sociology,Nankai University,Tianjin 300350)
出处
《心理与行为研究》
北大核心
2025年第2期233-240,共8页
Studies of Psychology and Behavior
基金
教育部产学合作协同育人项目(231000287155910)
国家社科项目(19FSHB010)。
关键词
人工智能
道德判断
社会认知
拟人化
CNI模型
artificial intelligence
moral judgment
social cognition
anthropomorphism
CNI model
