关键词: 算法决策, 人类决策, 决策趋避, 心理动机, 人?算法交互

Abstract:

With the advantages of objectivity, accuracy, high speed and low cost, algorithmic decision-making has been widely used in human daily life, such as medical, judicial, recruitment and transportation situations. How will people react to the shift from traditional human decision-making to the newly emerged algorithmic decision-making? If people perceive algorithms as social actors, there would be no difference when faced with the same decision made by two different agents. However, researches show that algorithmic decision-making is more related to different responses in individuals than human decision-making on the same content. In other words, people will approach or avoid the same algorithmic decision-making, which is defined as the algorithmic decision-making approach and avoidance. Specifically, the algorithmic decision-making approach means that algorithmic decision-making is considered fairer, less biased, less discriminatory, more trustworthy, and more acceptable than human decision-making. But the algorithmic decision-making avoidance is the other way around. By analogy with the distinct ideologies when facing outgroup members, the process motivation model of algorithmic decision-making approach and avoidance simulates human psychological motivation when facing the same decisions made by algorithms and humans. Based on the premise that quasi-social interaction (relationship) and interpersonal interaction (relationship) develop parallel, the theory summarizes the three interaction stages between humans and algorithms. Namely, the interaction of initial behavior, the establishment of quasi-social relationships and the formation of identity. Furthermore, it elaborates how cognitional, relational, and existential motivation trigger individual approach and avoidance responses in each specific stage. More precisely, it occurs to meet the cognitive motivational needs to reduce uncertainty, complexity, and ambiguity in the interaction of the initial behavior stage, fulfill the relational motivational needs for establishing belonging and social identity in the establishment of the quasi-social relationship stage, and to satisfy the motivational needs for coping with threats and seeking security in the identity formation stage. In accordance with the three psychological motivations of cognition, relationship, and existence, the process motivational theory introduces six influencing factors, such as cognitive load, decision transparency, moral status, interpersonal interaction, reality threat and identity threat respectively. For future directions, we suggest that more researches are needed to explore how mind perception and intergroup perception influence algorithmic decision-making approach and avoidance. Meanwhile, what is the reversal process of the algorithmic decision-making approach and avoidance from a social perspective and what other possible motivations are associated with it are also worthy of consideration.

Key words: algorithmic decision-making, human decision-making, decision-making approach and avoidance, mental motivation, human-algorithm interaction