Artificial intelligence (AI) is increasingly being explored as a means of supporting social care; however, limited public understanding of how these systems operate presents challenges for responsible implementation. This project develops an interactive web-based application designed to support non-expert understanding of AI within public-facing settings.

The system enables users to explore key aspects of AI through a sequence of short, interactive activities that demonstrate how outputs are generated and where limitations can arise. Built using a full-stack architecture, the application integrates a React frontend with a Node.js backend and external AI services to generate real-time responses. Design decisions prioritised accessibility, clarity, and reliability to ensure usability within short event-based interactions.

Evaluation findings indicate that users were able to interact effectively with the system and demonstrated increased awareness of issues such as bias, fairness, and reliability. However, results also show that users do not always interpret AI behaviour as intended, particularly when outputs deviate from expectations or lack sufficient explanation.

Overall, the project demonstrates that short, interactive experiences can support more informed public understanding of AI, while also showing that user-facing systems must communicate their behaviour and limitations clearly.