Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

InterFlow: Designing Unobtrusive AI to Empower Interviewers in Semi-Structured Interviews

Yi Wen, Yu Zhang, Sriram Suresh, Zhicong Lu, Can Liu, Meng Xia

TL;DR

InterFlow addresses the challenge of conducting high-quality semi-structured interviews by providing an AI-powered visual scaffold that adapts the interview script and supports real-time data sensemaking. The system integrates three modes of information capture (manual tagging, AI-assisted summary, and a co-interviewer agent) within a lightweight, unobtrusive interface comprising an interactive script, a visual timer, and a mixed-initiative capture module. In a within-subject study (N=12), InterFlow reduced interviewers' cognitive load and improved script navigation and information capture, though the actionability of AI suggestions varied with conversational dynamics and raised concerns about distraction and authenticity. The paper derives design implications for unobtrusive AI in time-sensitive, attention-demanding tasks and discusses limitations and avenues for future work, including data privacy and robustness.

Abstract

Semi-structured interviews are a common method in qualitative research. However, conducting high-quality interviews is challenging, as it requires interviewers to actively listen to participants, adapt their plans as the conversation unfolds, and probe effectively. We propose InterFlow, an AI-powered visual scaffold that helps interviewers manage the interview flow and facilitates real-time data sensemaking. The system dynamically adapts the interview script to the ongoing conversation and provides a visual timer to track interview progress and conversational balance. It further supports information capture with three levels of automation: manual entry, AI-assisted summary with user-specified focus, and a co-interview agent that proactively surfaces potential follow-up points. A within-subject user study (N = 12) indicates that InterFlow reduces interviewers' cognitive load and facilitates the interview process. Based on the user study findings, we provide design implications for unobtrusive and agency-preserving AI assistance under time-sensitive and cognitively demanding situations.

InterFlow: Designing Unobtrusive AI to Empower Interviewers in Semi-Structured Interviews

TL;DR

InterFlow addresses the challenge of conducting high-quality semi-structured interviews by providing an AI-powered visual scaffold that adapts the interview script and supports real-time data sensemaking. The system integrates three modes of information capture (manual tagging, AI-assisted summary, and a co-interviewer agent) within a lightweight, unobtrusive interface comprising an interactive script, a visual timer, and a mixed-initiative capture module. In a within-subject study (N=12), InterFlow reduced interviewers' cognitive load and improved script navigation and information capture, though the actionability of AI suggestions varied with conversational dynamics and raised concerns about distraction and authenticity. The paper derives design implications for unobtrusive AI in time-sensitive, attention-demanding tasks and discusses limitations and avenues for future work, including data privacy and robustness.

Abstract

Semi-structured interviews are a common method in qualitative research. However, conducting high-quality interviews is challenging, as it requires interviewers to actively listen to participants, adapt their plans as the conversation unfolds, and probe effectively. We propose InterFlow, an AI-powered visual scaffold that helps interviewers manage the interview flow and facilitates real-time data sensemaking. The system dynamically adapts the interview script to the ongoing conversation and provides a visual timer to track interview progress and conversational balance. It further supports information capture with three levels of automation: manual entry, AI-assisted summary with user-specified focus, and a co-interview agent that proactively surfaces potential follow-up points. A within-subject user study (N = 12) indicates that InterFlow reduces interviewers' cognitive load and facilitates the interview process. Based on the user study findings, we provide design implications for unobtrusive and agency-preserving AI assistance under time-sensitive and cognitively demanding situations.
Paper Structure (81 sections, 2 equations, 10 figures, 4 tables)

This paper contains 81 sections, 2 equations, 10 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. AI Support in Qualitative Research
  4. Designing Unobtrusive AI in Attention-Intensive Tasks
  5. Formative Study
  6. Background and Purpose
  7. Participants and Procedure
  8. Analysis
  9. Findings
  10. [F1] Difficulties with navigating the interview script
  11. [F2] Interrupted conversation due to manual note-taking
  12. [F3] Lack of awareness of time allocation
  13. [F4] Collaborative interviewing: two interviewers for workload sharing
  14. Design Goals
  15. [G1] Help Interviewers Manage the Interview Flow
  16. ...and 66 more sections

Figures (10)

  • Figure 1: (a) Pre-interview setup page for uploading script and contextual information (research question, background introduction, planned interview time) (b) Enhanced interview script with visual hierarchy.
  • Figure 2: Mixed-initiative Information Capturing
  • Figure 3: Other Examples of the Co-interviewer Agent's Suggestions
  • Figure 4: System Data Flow
  • Figure 5: Baseline system contains a chat panel (left) powered by OpenAI Realtime API and a text editor (right). Before the interview starts, users can provide prompts to configure the AI assistant. The AI assistant will listen to the conversation and proactively provide help in text format during the interview.
  • ...and 5 more figures