Algorithm acceptance in COVID-19-related decision making

In times of the COVID-19 pandemic, difficult decisions such as the distribution of ventilators must be made. For many of these decisions, humans could team up with algorithms; however, people often prefer human decision-makers. We examined the role of situational (morality of the scenario; perspective) and individual factors (need for leadership; conventionalism) for algorithm preference in a preregistered online experiment with German adults (n = 1,127). As expected, algorithm preference was lowest in the most moral-laden scenario. The effect of perspective (i.e., decision-makers vs. decision targets) was only significant in the most moral scenario. Need for leadership predicted a stronger algorithm preference, whereas conventionalism was related to weaker algorithm preference. Exploratory analyses revealed that attitudes and knowledge also mattered, stressing the importance of individual factors.

在新冠病毒疫情肆虐之际，诸如呼吸机分配等艰难决策必须作出。对于众多此类决策，人类可以与算法协作；然而，人们往往更倾向于信任人类的决策者。本研究通过一项预先注册的在线实验，考察了情境因素（例如场景的道德性；视角）和个人因素（例如领导力需求；传统主义）对算法偏好所产生的影响，实验对象为1,127名德国成年人。正如预期，道德负担最重的场景中，算法偏好最低。视角（即决策者与决策目标）的影响仅在道德负担最重的场景中显著。领导力需求预测了更强的算法偏好，而传统主义则与较弱的算法偏好相关。探索性分析揭示了态度和知识同样重要，强调了个人因素的重要性。