“It seemed like an annoying woman”: On the Perception and Ethical Considerations of Affective Language in Text-Based Conversational Agents

Abstract

Previous research has found that task-oriented conversational agents are perceived more positively by users when they provide information in an empathetic manner compared to a plain, emotionless information exchange. However, users' perception and ethical considerations related to a dialog systems' response language style have received comparatively little attention in the field of human-computer interaction. To bridge this gap, we explored these ethical implications through a scenario-based user study. 127 participants interacted with one of three variants of an affective, task-oriented conversational agent, each variant providing responses in a different language style. After the interaction, participants filled out a survey about their feelings during the experiment and their perception of various aspects of the chatbot. Based on statistical and qualitative analysis of the responses, we found language style played an important role in how humanlike participants perceived a dialog agent as well as how likable. Language style also had a direct effect on how users perceived the use of personal pronouns 'I' and 'You' and how they projected gender onto the chatbot. Finally, we identify and discuss ethical implications. In particular we focus on what factors/stereotypes influenced participants' impressions of gender, and what trade-offs a more human-like chatbot brings.

Increased morality through social communication or decision situation worsens the acceptance of robo-advisors

“…The result for anthropomorphism is in line with our expectations of a higher perceived anthropomorphism for social communication considering prior research showing that social robot interactions (Fraune, Oisted, et al, 2020), self-disclosure (Jolley, 2019;Audet and Everall, 2010), content intimacy (Cai et al, 2022;Seeger et al, 2021;Catania et al, 2020), and personal pronouns (Vanderlyn et al, 2021;Seeger et al, 2021;Large et al, 2019;Aggarwal and McGill, 2007) increase anthropomorphism. Despite different levels of anthropomorphism, subjects did accept the robot's recommendation equally often in both communication styles.…”

“…One study found that the "we" should be emphasized when communicating hospital decisions (Pfaff and Braithwaite, 2020). Another study found that the use of personal pronouns in voice agents contributes to a more personal and natural user experience (Vanderlyn et al, 2021).…”

“…If we-phrases contribute to a stronger ingroup perception, this could buffer algorithm aversion since Fraune, Šabanović, and Smith, 2020 found that ingroup robots are preferred over outgroup humans. Furthermore, the use of self-referential pronouns increases the anthropomorphism of voice agents (Seeger et al, 2021;Vanderlyn et al, 2021;Large et al, 2019) or cars (Aggarwal and McGill, 2007), the likability of voice agents (Vanderlyn et al, 2021), and the competence perception of robots (Dautzenberg et al, 2021).…”

The Impact of Gender and Personality in Human-AI Teaming: The Case of Collaborative Question Answering

“…Personality has been found to offer consistency to the interaction [13,46], helping users feel that they are talking to only one person throughout the conversation [62]. Personality also improves the chatbot user experience [62] by enhancing conversational agents' likability and humanness, [65] as a pure information exchange gives way to a more empathetic and self-referencing language style, which is generally preferred and perceived as more realistic, [65] in particular when displaying agreeableness. As observed by [66], displaying humbleness, as well as friendliness, increased users' perceptions of personalisation and social presence, resulting in greater experience satisfaction.…”

“…Such effects imply that designers can learn to control, through chatbot personality, how users attribute characteristics to the CA, and use humanness to manage user's expectations and trust. [62] In some cases, specific personality models such as the Big Five, [62,55], Myers-Briggs [65] and DISC theory [32] are used to inform agent design decisions, which in turn determine specific dialogue choices. [54] [27] explored the impact of CA personality on teamwork using a collaborative gaming challenge where the agent displayed two Big Five personality traits, extraversion and agreeableness, utilizing both verbal and nonverbal cues.…”