Description: Qualitative interviewing is a widely used research method that is very effective for gaining insights into the target user group. However, they are very time-consuming, which typically results in small sample sizes, and may lead to lack of diverse perspectives. The rise of GenAI and embodied AI agents provides a great opportunity to support researchers with this task. Given the complexity of interviews and taking difference in how humans conduct interviews into account, embodied AI agent (i.e., human avatars) may allow further standardization of the method. Moreover, it can enable research to systematically tackle biases towards the interviewer.

Goal of Thesis

The goal of this Bachelor/Master thesis (level will determine the scope) is to investigate how different designs of embodied AI-Agents affect participants in survey interviews.

1. Review of the related literature of interview techniques, biases (sex, gender, ethnicity etc.) and existing systems in conducting qualitative interviews through AI (e.g., interview chatbots / appareance / social cues).
2. Formulate hypotheses, design solutions, and develop prototypes based on the related literature.
3. Conduct an evaluation to assess how different designs of embodied AI-Agents impact participants in qualitative interviews.

Requirements

• Experience with LLMs, C# or Java; ideally, but not required, Unity3D / animation
• Strong interest in embodied AI agents in Virtual Reality
• You have read the following papers: [1] [2]

                        Optional: different types of interviews, but shows bias [3]