Mitigating Values Debt in Generative AI: Responsible Engineering with Graph RAG

Generative AI technologies are rapidly transforming industries such as healthcare, education, and transportation. However, this progress often incurs a Values Debt-ethical and operational deficits due to insufficient ethical considerations during development. This paper examines Values Debt in Gen-erative AI and introduces the Helpful, Honest, Harmless (HHH) framework to align AI systems with human values. In developing GRAISE, a Graph RAG-based chatbot for aviation safety, the HHH framework is applied to integrate ethical practices through-out the development process. This case study demonstrates how the HHH framework addresses ethical challenges and provides reliable contextual information to enhance pilot communication, exemplifying responsible AI engineering. These findings advocate for the broader adoption of ethical AI frameworks across various sectors, promoting trust and integrity in AI applications.