Moral Machines: Identifying Cues of Humanness that Influence Judgments of Robot Morality

近年来，关于决策机器的道德心理学的研究在一系列学科中有了显着的增长。这并不奇怪，因为许多重要的决策权已经在各种道德相关的背景下委托给了机器（例如，医院分诊协议）。然而，尽管有可能产生重大的社会效益，但仍然有相当多的人不愿意接受道德决策机器。出于这个原因，研究人员试图更好地了解这种态度背后的心理学。 公众对机器决策者的信任的一个关键组成部分是人们在心理上将机器类比为人类的程度。事实上，HRI研究的一个日益增长的焦点是创造显示“社会智能”的机器，即，可以以人类认为无缝和吸引人的方式与人类互动。为此，我和我的研究团队，以及多伦多大学机械与工业工程系自主系统和生物机电实验室的合作者，正在提议操纵机器人的心理特征，这些心理特征会影响人类在三种道德决策背景下（牺牲困境，资产分配和囚犯困境情景）对机器人人性的看法。为此，我的学生、同事和我将呼吁最新的心理学模型，沿着心理人性的四个维度来操纵机器人交互：（1）自主性（2）社会联系（3）自我意识情感（4）价值观/道德原则。在拟议的研究中，参与者将与一个机器人互动，该机器人已被编程为沿着一个或多个维度随机变化。我完全有能力进行拟议的研究。我是一名实验社会心理学家，在道德心理学方面有20年的出版物。虽然大部分关于人机交互的研究都是由机械工程师和计算机科学家进行的，但我可以提供一个关于人机界面“人”方面的最新观点。此外，多伦多大学拥有工程师和心理学家之间合作的独特机制。我是心理学和工程合作专业的创始成员，心理学系和土木与矿物工程，电气与计算机工程系的硕士/博士课程。机械与工业工程该计划促进了一些工程心理学研究合作，如人为因素和心理学设计。我目前在机械和工业工程（MIE）的四名学生的博士委员会任职。来自Goldie Nejat博士实验室的MIE学生是我实验室的积极参与者。

Recent years have seen marked growth in research across a range of disciplines on the moral psychology of decision-making machines. This is not surprising given that many important decision-making powers have been delegated to machines in a variety of morally-relevant contexts (e.g., hospital triage protocols). There remains, however, considerable popular reluctance to embrace moral decision making machines, despite the possibility of significant societal benefits. For this reason, researchers have sought to gain a greater understanding of the psychology underlying such attitudes. One critical component of the public's trust in machine decision makers is the degree to which people mentally analogize the machine to a human. Indeed, a growing focus in HRI research is on creating machines that display `social intelligence,' i.e., can interact with humans in ways that humans find seamless and engaging. To that end, my research team and I, with collaborators at the University of Toronto Autonomous Systems and Biomechatronics Lab in the Department of Mechanical and Industrial Engineering, are proposing to manipulate psychological features of robots that influence humans' perceptions of the robots' humanness in three moral decision contexts (sacrificial dilemmas, asset allocation, and prisoner's dilemma scenarios). To do so, my students, colleagues, and I will call upon recent psychological models to manipulate robot interactants along four dimensions of psychological humanness:  (1) Autonomy (2) Social Connection, (3) Self-Aware Emotions, and (4) Values/Moral Principles. In the proposed studies, participants will interact with a robot that has been programmed to vary randomly along one or more of these dimensions. I am well-positioned to conduct the proposed research. I am an experimental social psychologist with 20 years of publications in moral psychology. Whereas much of the research on human-robot interaction has been conducted by mechanical engineers and computer scientists, I can offer an up-to-date perspective on the `human' side of the human-robot interface. In addition, the University of Toronto possesses a unique mechanism for collaboration between engineers and psychologists. I am a charter member of the Psychology and Engineering collaborative specialization, a Masters/PhD program of the Department of Psychology and the Departments of Civil & Mineral Engineering, Electrical & Computer Engineering. and Mechanical & Industrial Engineering. This program has spurred several Engineering-Psychology research collaborations in areas such as human factors and psychology-informed design. I currently serve on PhD committees of four students in Mechanical and Industrial Engineering (MIE). One MIE student from the laboratory of Dr. Goldie Nejat is an active participant in my laboratory.