This paper presents a new experimental paradigm for studying human-computer interaction, showing that interactions with computers are fundamentally social. Five experiments demonstrate that users exhibit social behaviors toward computers, not due to beliefs that computers are human or human-like, but because of the social cues they provide. These cues include language output, responses based on prior inputs, role-playing, and human-sounding voices. The studies show that users apply social rules to computers, such as politeness norms, notions of self and other, gender stereotypes, and social attributions. These responses are automatic and unconscious, and are not due to ignorance, dysfunction, or belief that users are interacting with programmers. The results suggest that social responses to computers are common and easy to generate, and have implications for design and usability testing. The findings indicate that human-computer interaction is social-psychological, and that social cues are powerful and can be used to create social responses. The research also suggests that low-overhead agents can be easily produced and can generate a wide range of social responses, and that concerns about photorealistic representations may be overrated. The studies highlight the importance of social cues in human-computer interaction and suggest that design should consider these cues to create more effective and socially responsive interfaces.This paper presents a new experimental paradigm for studying human-computer interaction, showing that interactions with computers are fundamentally social. Five experiments demonstrate that users exhibit social behaviors toward computers, not due to beliefs that computers are human or human-like, but because of the social cues they provide. These cues include language output, responses based on prior inputs, role-playing, and human-sounding voices. The studies show that users apply social rules to computers, such as politeness norms, notions of self and other, gender stereotypes, and social attributions. These responses are automatic and unconscious, and are not due to ignorance, dysfunction, or belief that users are interacting with programmers. The results suggest that social responses to computers are common and easy to generate, and have implications for design and usability testing. The findings indicate that human-computer interaction is social-psychological, and that social cues are powerful and can be used to create social responses. The research also suggests that low-overhead agents can be easily produced and can generate a wide range of social responses, and that concerns about photorealistic representations may be overrated. The studies highlight the importance of social cues in human-computer interaction and suggest that design should consider these cues to create more effective and socially responsive interfaces.