Project Information
- Category: Machine Learning & Conversational AI
- Date: 2025
- Technologies: Django, Retell AI, Prolific, Python
- Role: AI Engineer & Full-Stack Developer
Voice AI Note-Taking Study
A research project studying how an AI agent behaves in a voice-based hiring workflow across three note-taking conditions, using Retell AI to simulate realistic call scenarios. Data collection is currently in progress.
Experimental Design:
- Condition 1: No AI note-taking — baseline agent behavior with no memory assistance.
- Condition 2: AI note-taking enabled — agent has access to AI-generated call notes.
- Condition 3: Alternate note setup — a second baseline for comparison.
Key Behaviors Tracked:
- Memory Retention: Whether the agent retains key details across turns without re-asking or contradicting itself.
- Information Handling: Whether the agent asks for missing information vs. guessing.
- Decision Consistency: Whether the agent follows hiring decision rules consistently across scenarios.
- Adaptability: How the agent recovers when the customer introduces new or conflicting requirements.
Data collection is currently in progress. Participants are being recruited via Prolific, and Retell's standardized customer role ensures scenario consistency across all three conditions.