Perceiving artificial social agents

摘要

Three experiments were conducted to examine perceptions of a natural language computer interface (conversation bot). Participants in each study chatted with a conversation bot and then indicated their perceptions of the bot on various dimensions. Although participants were informed that they were interacting with a computer program, participants clearly viewed the program as having human-like qualities. Participants agreed substantially in their perceptions of the bot’s personality on the traits from the five-factor model (Experiment 1). In addition, factors that influence perceptions of human personalities (e.g., whether one uses another’s first name and response latency) also affected perceptions of a bot’s personality (Experiments 2 and 3). Similar to interactions with humans, the bot’s perceived neuroticism was inversely related to how long individuals chatted with it.