Human-AI Conversational Role Reversal for Fraud Awareness: Interaction Design and Usability Evaluation with Older Adults

The rapid growth of information technology and social media has increased online scams targeting older adults. Existing interactive digital interventions remain limited, as older adults often struggle to understand complex scams and respond to evolving scenarios. To address this, we developed a generative AI-based role-reversal application in which the AI simulates a scam victim, while older adults act as persuaders to guide the AI in recognizing and avoiding fraud. A total of 40 older adults were recruited and assigned to either the experimental group (n = 20) or the control group (n = 20). Results showed positive user experience and improved self-efficacy in AI interaction. The experimental group reported significantly higher attractiveness, stimulation, and novelty. Significant group × age interactions were found for dependability and perspicuity, while group × education interactions affected stimulation and perspicuity, indicating moderating effects of demographic factors. These findings inform design guidelines for GAI-based fraud prevention and highlight future directions.