USING GENAI CHATBOTS: USER TYPOLOGY AND STRATEGIES TO INCREASE USER TRUST

The questionnaire used to evaluate user perceptions of a generative AI conversational agent (CgAI). Respondents assessed the agent's performance across various dimensions using a 5-point Likert scale, where 1 = strongly disagree and 5 = strongly agree.
The questionnaire items are categorized by conceptual dimensions to facilitate interpretation, although they were presented in randomized order during data collection. Each item is preceded by a code that corresponds to variables listed in Tables 2, 3, and 4 of the main article.
The dimensions and corresponding items are as follows:
Performance
Measures how effectively the CgAI agent aids users in completing tasks.
PER1–PER5: Evaluate efficiency, speed, ease of interaction, and task support.
Security
Assesses user perceptions of data privacy and trust in secure handling.
SEC1–SEC4: Gauge trust in data safety, privacy respect, and secure communication.
Beneficence
Reflects the agent’s intent and competency in providing helpful, trustworthy information.
BEN1–BEN4: Measure perceived helpfulness, reliability, and user-centric behavior.
Anthropomorphism
Captures the degree to which the agent is perceived as understanding and responsive like a human.
ANT1-ANT4: Evaluate perceived understanding, contextual awareness, and adaptability to feedback.
Immediacy
Refers to the personalization and continuity of user experience over time.
IMM1–IMM3: Assess memory of past interactions, adaptation to preferences, and tailored suggestions.
Cognitive-Based Trust
Relates to the perceived accuracy and consistency of the agent’s responses.
CBT1–CBT3: Measure factual reliability and error rates.
Affect-Based Trust
Reflects emotional trust, including comfort, confidence, and emotional sensitivity.
ABT1–ABT4: Evaluate emotional support, confidence-building, and responsiveness to emotional cues.

本问卷用于评估用户对生成式AI对话智能体（generative AI conversational agent, CgAI）的感知评价。受访者通过5级李克特量表（5-point Likert scale）对该智能体的多维度表现进行评分，其中1代表“完全不同意”，5代表“完全同意”。
问卷题目按概念维度进行分类以便于结果解读，但在数据收集阶段题目呈现顺序为随机化。每道题目前均附有代码，对应主文章中表2、表3及表4所列的变量。
各维度及对应题目如下：
性能（Performance）：衡量该生成式AI对话智能体协助用户完成任务的有效程度。PER1–PER5：评估交互效率、响应速度、交互便捷性及任务支持度。
安全性（Security）：评估用户对数据隐私的感知以及对智能体安全处理数据的信任度。SEC1–SEC4：衡量用户对数据安全、隐私保护及安全通信的信任程度。
行善性（Beneficence）：反映智能体提供有益、可信信息的意图与能力。BEN1–BEN4：衡量用户感知到的有用性、可靠性及以用户为中心的行为表现。
拟人化（Anthropomorphism）：捕捉用户感知中智能体类人般的理解能力与响应性程度。ANT1–ANT4：评估智能体的感知理解能力、情境意识及对反馈的适应性。
即时性（Immediacy）：指随时间推移的用户体验个性化与持续性。IMM1–IMM3：评估智能体对过往交互的记忆能力、对用户偏好的适配性及个性化建议输出。
基于认知的信任（Cognitive-Based Trust）：关联用户对智能体回复准确性与一致性的感知。CBT1–CBT3：衡量信息的事实可靠性与错误率。
基于情感的信任（Affect-Based Trust）：反映情感层面的信任，包括舒适感、信心及情感敏感度。ABT1–ABT4：评估情感支持、信心构建及对情感线索的响应性。