Aug 2026· Communication and Change· Vol 2· 0 citations· 69 references
TL;DR
Investigating how chatbots’ presented gender, together with users’ gender and gender traits, influence users’ perceptions of chatbot credibility and social attraction revealed that users’ gendered self-concept modulates their perceptions of the chatbot.
Abstract
As conversational agents become a core interface in service, education, and decision support, understanding how users’ self-concept shapes their interaction with gendered AI agents is critical. This study investigates how chatbots’ presented gender, together with users’ gender and gender traits, influence users’ perceptions of chatbot credibility and social attraction. Across two experimental studies (N₁ = 250, N₂ = 228), participants engaged with a chatbot presented as female, male, or gender-neutral. Study 1 employed a rule-based chatbot, whereas Study 2 used a generative AI chatbot for a more naturalistic interaction. In Study 1, results revealed that users’ gendered self-concept modulates their perceptions of the chatbot, particularly when chatbot gender aligns or misaligns with participants’ self-perceived masculinity or femininity. The more masculine users were, the more they perceived the male chatbot as competent; by contrast, users with higher femininity evaluated the neutral chatbot more favorably in goodwill, trustworthiness, and social attraction. However, these effects were not observed in Study 2, suggesting that advanced AI performance may dilute social identity-based biases. We further discuss how interaction effects reflect broader sociocultural schemas embedded in digital communication and highlight ethical implications for AI design, including risks of reinforcing gender stereotypes and marginalizing non-binary identities. Findings underscore the need for inclusive, identity-aware chatbot design that balances personalization with mindfulness.
ChatGPT, a chatbot powered by a large language model (LLM), is used by millions globally. Whether it is perceived as an anthropomorphic agent is currently being mapped out. Notably, the degree and nature of how ChatGPT is gendered remains unclear, despite agent gender cues being known to impact user attitudes and behaviors. ChatGPT, with its natural communication style and minimal gender cues, may offer radical insights on agent gendering. In this online mixed methods study, we invited first-time and experienced users to undertake tasks with ChatGPT. We surveyed their perspectives at two time-points following the interaction. Results indicated that ChatGPT’s communicative abilities correlated with gendered traits and its perceived capabilities reflected gender stereotypes. Our work suggests that ChatGPT is not gender-neutral, with implicit gender stereotyping occurring in the absence of gender cues and regardless of exposure. We advocate for conscientious design interventions to overcome agent-based perpetuation of human stereotypes.
Madeleine Steeds, Katie Seaborn, Takao Fujii et al.· International Conference on...· 0 citations
Manipulative interactions were associated with evaluative contexts and steering behaviors, whereas inappropriate interactions were linked to relational contexts and social discomfort, and implications for trustworthy AI design and for regulatory frameworks such as the EU AI Act are discussed.
Vanessa Budde, Eran Toch· Message Understanding Confer...· 0 citations
We investigate how gender affects social interaction outcomes, and to what extent gendered language mediates the association between gender and interpersonal liking. From a multimodal corpus of dialogue interactions in which participants rated and estimated likeability (CANDOR), we use inferential statistical analyses to identify expressions associated with gender, likeability, and the liking gap, the difference between how much an interlocutor likes another and how liked interlocutors imagine they are. Results reveal a Dunning Kruger effect of likeability, as well as evidence that an interaction of gender and liking is partly explained by gendered language independent of speaker gender. These findings bridge and expand previous work on liking, gender attitudes, gendered language, and likeable language. Our results suggest that aspects of stereotypically feminine language can enhance likeability regardless of the speaker’s gender identity.
Teona Banu, Carl Vogel· Dialogue and Discourse· 0 citations
The study found that although most participants realized that AI Chatbots did not have human emotions at the cognitive level and only relied on Algorithms and databases to answer, they often still experienced a real sense of mutual intimacy.
In academic settings, communicative competence is a foundation stone of successful interactions as it brings learner’s linguistic and sociocultural awareness into prominence. The current research investigates the request strategies applied by Kurdish EFL learners in their emails to instructors, with an emphasis on gender-based interaction patterns and strategies of politeness. The study employs a mixed method approach to explore 126 authentic email requests gathered from male and female undergraduate learners. The Cross-Cultural Speech Act Realization Project (CCSARP) was adopted to evaluate the emails. This framework was developed by Blum-Kulka et al., highlighting modifiers and levels of directness. The results demonstrate moderate level of pragmatic awareness among students, based on strategies of direct request, particularly among male learners, whereas female students preferred conventionally indirect approaches, signaling a stronger movement towards politeness. The study reveals notable gender-related differences in crafting requests, with female students showing more significant variability in the use of strategies. All of these provide strong justification for the integration of pragmatic instruction in EFL programs intended to foster students’ communicative skills.
S. Barzani, R. Ahmad, F. R. Omar· Journal of ethnic and cultur...· 0 citations
Mental health challenges among university students have become increasingly prevalent, while existing digital interventions often struggle to sustain user engagement and provide meaningful emotional support. Although previous studies have proposed various explanations for the limited acceptance of AI-based mental health chatbots, the role of affective experience remains insufficiently understood. This study proposes an affect-centered framework to examine how different virtual agent persona styles (Listener, Guide, and Motivator) influence university students’ psychological acceptance of mental health chatbots.
Drawing on the Behavior Change Wheel (BCW) and the Unified Theory of Acceptance and Use of Technology (UTAUT), this study developed a dual-pathway model to investigate the mediating role of affective engagement in AI-based mental health chatbot acceptance. Data were collected from 266 university students who were randomly assigned to scenario-based chatbot stimuli representing three virtual agent persona styles. Partial least squares structural equation modeling (PLS-SEM), artificial neural network (ANN) analysis, and partial least squares multi-group analysis (PLS-MGA) were conducted to examine the proposed relationships and persona-related differences.
The results indicate that most proposed hypotheses were supported. Affective engagement emerged as a central mechanism linking persona-related characteristics and psychological acceptance. Multi-group analysis further revealed that although the overall acceptance mechanism remained largely consistent across the three persona conditions, significant differences existed in the strength of key relationships. Specifically, the effect of affective engagement on behavioral intention was stronger in the Guide and Motivator conditions than in the Listener condition.
These findings suggest that virtual agent persona styles do not fundamentally alter the psychological acceptance mechanism but can influence the strength of critical affective pathways. The study highlights the importance of affective engagement in AI-supported mental health interactions and provides practical implications for designing more adaptive, emotionally responsive, and user-centered mental health chatbots.
Qi Wu, Shi-Zhu Lu, Yu-Qi Yan et al.· Frontiers in Psychology· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
