Table 2.

Expert interview item analysis.

Module	Sub-dimension	Question No.	Interview question (semi-structured guidance)	Purpose and explanation
I. Emotional design	Visceral level	Q1-1	What key features do you think an AI virtual companion should possess in terms of appearance or interface to create an immediate good impression on single young adults?	To deepen the definition of “visual appeal” and capture design guidelines.
		Q1-2	What experience of success or failure can you share to avoid the “uncanny valley” effect?	To gather feasible design boundaries and risks.
	Behavioral level	Q1-3	During interactions, which details can best enhance the “dialogue fluency” and the “sense of being understood”? Please provide real-life examples.	To clarify the evaluation dimensions of “interaction quality.”
		Q1-4	How do you measure the impact of AI response speed and semantic accuracy on emotional experience?	To provide measurement ideas for questionnaire indicators.
	Reflective level	Q1-5	How can AI companions provide users with emotional support or self-recognition through symbols, stories, or a sense of ritual?	To explore mechanisms for deep emotional resonance.
II. Personalization and intelligent adaptability	Data-driven learning	Q2-1	What user data do you think the system should record and learn from to truly provide a personalized companionship experience?	To delineate the scope of data collection.
		Q2-2	When an AI companion “misremembers” or “forgets,” what are the typical emotional responses of users?	To reveal the risks of adaptive failures.
	Contextual adaptation	Q2-3	How should an AI companion distinguish different emotional states (e.g., loneliness, anxiety, and excitement) in its response strategies?	To refine the logic of emotion recognition and feedback.
III. Technology perception and usage barriers	Perceived usefulness/ease of use	Q3-1	What specific pain points do single young adults most expect AI companions to address in daily life?	To clarify the scenarios of “perceived usefulness.”
		Q3-2	In actual deployments, which technical flaws are most likely to cause loss of users? How should they be improved in priority?	To rank improvement priorities.
IV. Ethical and social acceptance	Privacy security	Q4-1	What kind of design do you think should be made at the product, protocol, or operational level to make users believe that their chat content is absolutely confidential?	To correspond with the “trustworthiness” item in the questionnaire.
	Emotional dependency	Q4-2	How do you judge whether users are overly dependent on AI companions or not? What preventive mechanisms can be designed?	To enrich ethical indicators.
	Social image	Q4-3	How do you view the “stigma surrounding virtual companions”? What positioning or promotional strategies can help reduce negative labels?	To guide the setting of contextual variables.
V. Usage scenarios and cultural differences	Scenario triggers	Q5-1	In your observations, what time, device, or location do users interact with AI companions most frequently?	To define “high-frequency usage scenarios.”
	Cultural factors	Q5-2	Which kind of emotional design element should be particularly adjusted in different cultural or regional backgrounds? Please provide examples.	To provide a basis for cross-cultural extrapolation.
VI. Future trends and business strategies	Product iteration	Q6-1	What are the development trends in emotional design for AI companions over the next 3–5 years?	A predictive question for discussion.
	Business models	Q6-2	Among paid subscriptions, value-added services, or advertising placements, which model is more acceptable to single young adults?	To link to the derivative variables of “behavioral intention.”