厌恶算法还是欣赏算法?--人工智能时代的算法认知差异与算法信任建构  

Algorithm Aversion or Algorithm Appreciation?Algorithm Cognitive Differences and Algorithm Trust Construction

作　　者：杜严勇 DU Yanyong

机构地区：[1]同济大学人文学院

出　　处：《哲学分析》2022年第3期151-165,199,共16页Philosophical Analysis

基　　金：2020年国家社科基金重大项目“人工智能伦理风险防范研究”(项目编号:20&ZD041)阶段性成果。

摘　　要：尽管算法在精确性方面已经超过了人类,但不少人还是倾向于拒绝接受算法推荐与算法决策,产生这种“厌恶算法”现象的主要原因有人类认知特征与算法本身具有的局限性等。与此同时,也存在“欣赏算法”现象,即有人更愿意接受来自算法的推荐而不是人类的建议,其原因可能主要在于算法所具有的某些优越性。可以从技术、用户与人机互动三个维度着手建构合理的算法信任,进而避免智能时代的数字鸿沟,使算法更好地为人类服务。Although human has surpassed by algorithm in accuracy,many people still tend to refuse algorithmrecommendation and decision-making,which known as algorithm aversion.This essay argues that the main reasonsfor this phenomenon include human cognitive characteristics and the limitations of the algorithm itself.At the sametime,some people adhere more to advice when they think it comes from an algorithm than from a person,whichknown as an algorithm appreciation.We should construct reasonable algorithm trust from the three dimensions oftechnology,user and human-machine interaction,in order to avoid the digital divide in the intelligent era.

关 键 词：厌恶算法 欣赏算法 算法信任 

分 类 号：B80[哲学宗教—思维科学]