This lightning talk presents insights from a pilot program in an IT Faculty, where custom generative AI bots were integrated into online forums across 20 courses over two semesters in 2024. The AI bots were trained on specific course content and past student questions to provide tailored responses to student inquiries, with all responses reviewed by teaching staff before being released to students.

This approach, distinct from the direct use of large language models (LLMs) like ChatGPT or Claude, offers targeted information aligned with course material and ensures accuracy while preventing the disclosure of assignment answers. The mechanism is designed to support large computer science courses, including first-year courses with over 1,000 students, where timely and comprehensive staff responses can be challenging.

This talk will explore the benefits and drawbacks of using generative AI bots in the CS context. It will also examine the factors influencing staff acceptance and trust in chatbot responses. Key lessons learned and challenges encountered during the program’s implementation will also be shared.