03: Crafting Trustworthy Personalities

Introduction

The chapter discusses the importance of crafting trustworthy personalities for conversational user interfaces (CUIs), emphasizing that users naturally attribute personality traits to these systems, even when they know they are interacting with a machine. A well-defined personality enhances user trust and engagement, akin to the reliable and warm persona of Mr. Rogers, which fosters a sense of familiarity and safety. The text distinguishes between “personality,” which refers to the character traits and behaviors of the CUI, and “persona,” a standardized mental image inferred from the interface’s voice and language. It highlights the need for consistency in personality to avoid user confusion, contrasting effective personalities with erratic ones that create uncertainty. Additionally, the chapter underscores the ethical responsibility of designers to avoid stereotypes and ensure that the CUI’s prompts are structured thoughtfully, reflecting a careful balance between customization and consistency. This approach not only shapes user interactions but also influences the overall design and functionality of the conversational interface.

How a Personality Manifests

Personality in conversational user interfaces (CUIs) is shaped by three main elements: word choice, voice, and behavior. Users form perceptions of a CUI’s personality based on the language used in prompts, which can include specific jargon, emojis, and stylistic choices that convey character traits. For example, the Katy Perry chatbot uses energetic language and in-jokes to create a playful persona. The voice of the interface also plays a crucial role; studies suggest that users may judge personality more from vocal characteristics than from the actual content spoken. Additionally, the behavior of the CUI—such as politeness and responsiveness—contributes to the overall personality impression. Effective design integrates these elements to create a cohesive user experience, ensuring that the CUI’s identity is consistently communicated through its language, voice, and actions. This paper contributes theoretically by outlining how these components interact to shape user perceptions in human-computer interaction.

The Wrong Foundation for Personality

Many teams mistakenly view personality in conversational user interfaces as a superficial enhancement rather than a fundamental aspect of user experience. A well-defined personality is crucial for building user trust, and it requires more than just a collection of demographic traits or fun facts. For instance, creating a character like “Amy,” a half-Japanese virtual assistant with arbitrary interests, can lead to superficial and potentially offensive portrayals that fail to resonate with users. Instead, personality should be designed with actionable traits that influence dialogue and error-handling, rather than relying on demographics alone. This approach emphasizes the importance of understanding the target user group and their cultural context, as well as the ethical responsibility to avoid stereotypes and ensure inclusivity. Ultimately, effective personality design in conversational interfaces should focus on relatable character traits that enhance communication and user interaction, rather than merely reflecting demographic characteristics.

The Right Foundation for Personality

Designing an effective personality for a conversational user interface (CUI) involves a structured framework that emphasizes the relationship between the bot and the user, ensuring that the personality aligns with the interaction goals. Key components of this framework include defining interaction goals, determining the level of personification (how human-like the bot should be), understanding power dynamics in the user-bot relationship, selecting character traits that support interaction goals, establishing an appropriate tone, and outlining key behaviors for various scenarios.

The interaction goals serve as the foundation, guiding the design to ensure the personality enhances user experience rather than detracts from it. For instance, a bot designed for efficiency should avoid a quirky personality that could slow down interactions. The level of personification varies based on the task; simple transactional bots may require minimal personification, while personal assistants like Siri or Alexa benefit from a more developed character to foster trust and intimacy over time.

Power dynamics are crucial, as they dictate how the bot interacts with users based on their perceived roles. For example, a virtual hiring assistant must maintain a professional demeanor to respect the hierarchical relationship with HR managers. This understanding informs the bot’s language and behavior, ensuring it aligns with user expectations.

Character traits should be distinctive and relevant to the interaction goals, while tone—both in writing and voice—must reflect the bot’s personality and the context of the interaction. For example, a financial bot might adopt a formal tone, whereas a customer service bot could be more casual and warm.

Key behaviors outline how the bot should respond in various situations, such as handling errors or interruptions, ensuring consistency with its established personality. This framework not only aids in the design and development of the CUI but also fosters collaboration among team members and stakeholders, ensuring a cohesive approach to personality design.

The research and prototyping phases are essential for evaluating the effectiveness of the personality, with methods such as user interviews and testing to gather insights. Ethical considerations, including the implications of power dynamics and the need for transparency, are also vital in creating a responsible and trustworthy CUI. Overall, this framework provides a comprehensive approach to developing a conversational interface that effectively meets user needs while maintaining a coherent and engaging personality.

Avoid Racist Stereotypes

Race significantly influences perceptions and experiences in conversational interactions, particularly in the design of conversational AIs. A notable issue is the lack of representation for Black, Brown, and Indigenous voices, as most default voice assistants sound white. While some companies offer diverse voice options, they often still rely on stereotypes when defining virtual personalities. For instance, a case study revealed a chatbot designed for Spanish speakers was characterized by clichéd traits, such as being a Catholic father of five, which perpetuated harmful stereotypes and lacked cultural relevance.

Affinity bias plays a crucial role in user interactions, where individuals tend to prefer voices and avatars that resemble themselves. This bias can lead to exclusive designs that fail to represent diverse user groups adequately. Effective design requires thorough user research and collaboration with community members to avoid harmful stereotypes. The distinction between a bot’s likability and its effectiveness is essential; user testing should focus on task completion rather than mere preference. Ethical considerations in the design process are paramount, ensuring that the conversational user interface reflects genuine cultural understanding and avoids reinforcing negative stereotypes.

To Gender or Not?

The discussion emphasizes the intricate relationship between gender and personality in the context of conversational user interfaces (CUIs). It highlights that users often assign gender to synthetic voices, reflecting societal stereotypes and biases, which can lead to the reinforcement of sexism in design. The text defines gender as a social construct, distinct from biological sex, and stresses that it exists on a spectrum rather than a binary system. Designers are encouraged to consider whether a gender is necessary for a CUI, with examples like Capital One’s gender-neutral chatbot Eno and Planned Parenthood’s Roo, which educate users about gender while maintaining a non-binary identity. The research and prototyping phases are crucial for developing CUIs that resonate with diverse gender identities, as seen in the creation of a nonbinary synthetic voice by Andreea Danielescu, which was informed by the experiences of nonbinary individuals. The text advocates for ethical design practices that avoid stereotypes and promote inclusivity, urging designers to intentionally craft personalities based on traits rather than gender. This approach not only enhances user interaction but also aligns with social responsibilities in technology development.

To Avatar or Not?

The discussion centers on the use of avatars—visual representations of conversational user interfaces (CUIs)—and whether they enhance or hinder user interaction. While avatars can range from simple icons to realistic characters, the necessity of a visual representation is questioned. Many interactions occur without visual cues, such as phone calls, suggesting that avatars may not be essential for effective communication. The paper highlights the gendered nature of voice assistants like Alexa and Siri, noting their evolving responses to gender identity, which reflects broader societal changes. It emphasizes the importance of intentional design to avoid biases and the “uncanny valley” effect, where near-human avatars can evoke discomfort. The emotional impact of visual representations is significant; users may project past experiences onto avatars, complicating interactions. Ultimately, the paper advocates for usability testing with diverse groups to determine the effectiveness of avatars, while stressing that the quality of dialogue is paramount in establishing a distinct personality for CUIs. This contribution is primarily theoretical, offering insights into design considerations and the dynamics of human-computer interaction.

Consistency Versus Customization

A conversational user interface (CUI) should maintain a consistent personality to avoid confusing users, as mixed signals can lead to abandonment. However, customization may be necessary to cater to different user segments, such as age or cultural background. For instance, designers Phoebe Ohayon and Maikel van der Wouden created distinct personalities for a game targeting both teenagers and older users, finding that teens preferred a high-energy, directive style, while older users favored clarity and emphasis on key messages. This highlights the importance of adapting communication styles to enhance user experience without creating entirely different personalities. Ethical considerations arise when designing CUIs, as they must build trust across diverse user groups, ensuring that any adjustments remain within the same overarching personality framework. This approach allows for effective interaction dynamics, recognizing emotional and situational cues while maintaining a coherent identity.

The Last Word

The chapter emphasizes the importance of defining interaction goals and character traits for conversational user interfaces (CUIs) to ensure effective communication and relationship-building. Key points include understanding the necessary level of personification, power dynamics in interactions, and the tone that conveys trust and support. The personality of the CUI significantly influences user experience, shaping perceptions of trust, consistency, and respect. It is crucial to test these personality traits with diverse user groups to avoid biases and assumptions, particularly regarding sensitive topics like race and gender. The chapter also discusses the relationship between brand identity and CUI personality, highlighting the need for collaboration with branding teams to align the bot’s personality with brand values while avoiding overt sales tactics. The iterative process of prototyping and user research is essential for refining the CUI’s personality, ensuring it evolves effectively over time. Ethical considerations are underscored, stressing the responsibility to create a personality that does not cause harm. Overall, the chapter provides a theoretical framework for designing CUIs that resonate with users while maintaining a clear identity and purpose.