MotivationSocially interactive agents have now been developed and discussed for more than 20 years. While they have aptly been described as a testbed to gain a better understanding of human communication skills, they have also always been developed with a view to supporting users by means of helpful applications and implementations. In this regard, they have from the beginning raised interest from a psychological and ethical point of view: Here, it soon became apparent that early applications -even when they were not yet very sophisticated regarding their interaction abilities -already proofed to be able to elicit social reactions from the users. The goal of the chapter therefore is to summarize research on these social effects. While this has already been done from a psychological perspective (Krämer, Rosenthal-von der Pütten & Hoffmann, 2015), the combination of a psychological and an ethical view is unique to this chapter.Effects of socially interactive agents can be termed 'social' if a participant's emotional, cognitive, or behavioral reactions are similar to reactions shown during interactions with other human beings (for an early review see Krämer, 2005) or, in some respect, other animals (e. g. de Waal 2019). A number of studies demonstrate that those -actually inappropriate -reactions really do occur, sometimes even without the appearance of a humanlike character. Studies Nass et al., 1997;Nass et al., 1994)indicate that in interactions with computers politeness phrases are employed, principles of person perception and gender stereotypes apply, and liking is triggered in a similar way as within human relationships (computer that 'flatter' and give positive feedback are evaluated more positive). Although the studies suggest that a "rich human presentation" (Nass et al., 1994) is not necessary and specifically speech might be sufficient to trigger social reactions (Nass & Moon, 2000), studies suggest that intelligent virtual agents (IVA) might be especially prone to lead to social reactions. There is long-standing evidence that even subtle social phenomena such as impression management [see e.g. Leary, 1995] are prevalent in human agent interaction. When a human-like face is present, participants aim at leaving a favourable impression by e.g. choosing a socially desirable TV program (documentary about Albert Einstein compared to James Bond movie) (Krämer, Bente & Piesk, 2003) or by presenting themselves in a socially desirable way (Sproull et al., 1996). Krämer (2005) showed that IVAs affect the way in which users communicate with a TV-VCR system. When an IVA is visible instead of a graphical user interface or a user interface with speech output, users address the system significantly more often using natural speech rather than using a remote control. Additional qualitative analyses of the semantic content of all speech acts indicate that users seem to have a more human-like attitude and behavior towards the system when it is represented by an anthropomorphic agent. Similarly, Hoffmann, Krämer, Lam-Chi and...
