Abstract:With the rise of human-machine communication, machines are increasingly designed with humanlike characteristics, such as gender, which can inadvertently trigger cognitive biases. Many conversational agents (CAs), such as voice assistants and chatbots, default to female personas, leading to concerns about perpetuating gender stereotypes and inequality. Critiques have emerged regarding the potential objectification of females and reinforcement of gender stereotypes by these technologies. This research, situated in conversational AI design, aims to delve deeper into the impacts of gender biases in human-CA interactions. From a behavioral and communication research standpoint, this program focuses not only on perceptions but also the linguistic styles of users when interacting with CAs, as previous research has rarely explored. It aims to understand how pre-existing gender biases might be triggered by CAs' gender designs. It further investigates how CAs' gender designs may reinforce gender biases and extend them to human-human communication. The findings aim to inform ethical design of conversational agents, addressing whether gender assignment in CAs is appropriate and how to promote gender equality in design.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to explore the influence of gender bias in the interaction between humans and Conversational Agents (CAs). Specifically, the research focuses on the following aspects: 1. **How gender identity affects human - machine dialogue**: - Researchers hope to understand whether simply endowing Conversational Agents with gender identity will affect users' attitudes and their ways of communicating with CAs. - Specific questions include: Does the gender identity of CAs change users' communication attitudes and ways? 2. **Differences between gendered human - machine dialogue patterns and interpersonal dialogue patterns**: - The research will compare the dialogue patterns between gendered CAs and humans, as well as the differences in dialogue patterns between gendered humans. - The key question is: Is human - machine dialogue more neutral or more gender - biased than interpersonal dialogue? 3. **The influence of gendered human - machine dialogue on subsequent interpersonal dialogue**: - The research further explores whether communicating with gendered CAs will indirectly affect users' interpersonal communication in real life. - For example, if users are used to ordering female CAs as assistants, will this behavior pattern be transferred to their interactions with real women, resulting in disrespectful behavior? ### Research background With the development of human - machine interaction technology, more and more machines are designed with human - like characteristics, such as gender. However, this design may inadvertently trigger cognitive biases, especially gender biases. Currently, many Conversational Agents (such as voice assistants and chatbots) are mostly designed as female images, which has raised concerns about gender stereotypes and inequality. For example, UNESCO warns that this "gendered" technology may exacerbate the objectification of women, gender stereotypes, and gender power imbalances. ### Research methods To fully understand these problems, researchers propose a mixed - method research plan, including: - **Observational study**: Record natural conversations between human users and CAs with different gender voices in a laboratory environment and conduct qualitative analysis to detect gendered communication cues and patterns. - **Online experiment**: Quantitatively study the interactions between different gendered CAs (female, male, or neutral) and users, and evaluate users' expectations and evaluations of CAs. - **Text analysis**: Use natural language processing tools (such as LIWC) to analyze the conversation history between users and CAs, focusing on aspects such as emotion, tone, femininity, masculinity, and power dynamics. - **Laboratory experiment**: Test the changes in users' attitudes and behaviors after interacting with CAs with gender stereotypes (such as female assistants or technical experts), and observe how these changes affect subsequent conversations with human agents. ### Conclusion This research not only helps to understand the role of gender bias in human - machine dialogue but also provides ethical guidance for the future design of Conversational Agents. The research results will help us think about whether we should endow Conversational Agents with gender identity when designing them and how to avoid the continuation and intensification of gender bias.

Exploring Gender Biases in Language Patterns of Human-Conversational Agent Conversations

Exploration, detection, and mitigation: Unveiling gender bias in NLP

Does Gender Matter? Towards Fairness in Dialogue Systems

Gender Biases in Error Mitigation by Voice Assistants

Decoding Gender Representation and Bias in Voice User Interfaces (VUIs)

LGBTQ-AI? Exploring Expressions of Gender and Sexual Orientation in Chatbots

Caring for Communities through AI Language

Alexa, Google, Siri: What are Your Pronouns? Gender and Anthropomorphism in the Design and Perception of Conversational Assistants

Towards Investigating Biases in Spoken Conversational Search

Exploring Human's Gender Perception and Bias toward Non-Humanoid Robots

Learning from Red Teaming: Gender Bias Provocation and Mitigation in Large Language Models

Toward Gender-Inclusive Coreference Resolution: An Analysis of Gender and Bias Throughout the Machine Learning Lifecycle

Gender Bias in Online Language Translators: Visualization, Human Perception, and Bias/Accuracy Tradeoffs

Revealing Persona Biases in Dialogue Systems

A Survey on Gender Bias in Natural Language Processing

Gender Bias in AI Recruitment Systems: A Sociological-and Data Science-based Case Study

Understanding Public Perceptions of AI Conversational Agents: A Cross-Cultural Analysis

Unpacking the Black Box: Examining the (de)gender Categorization Effect in Human-Machine Communication

Mitigating Gender Bias in Natural Language Processing: Literature Review

Voices of Her: Analyzing Gender Differences in the AI Publication World

Public Perceptions of Gender Bias in Large Language Models: Cases of ChatGPT and Ernie