How to Explain Artificial Intelligence to Humans

QFD and AI share a common past, namely Transfer Functions that solve questions for the root cause of some desirable effect. In the future, QFD could help to explain AI. The paper examines similarities, it outlines where the significant differences are but also identifies the limitations of both AI and QFD. Both approaches do not have a feedback loop such as natural neural networks commonly have. However, since the Graph Model of Combinatory Logic models both AI and QFD, it is possible to embed both into a framework where processes, or programs, provide the necessary feedback loops. Finally, a proposal is presented on how to make AI explainable and acceptable to humans. We believe that training in QFD would help providers of AI-enabled products to better explain how AI works, using examples that are much easier to communicate to regulators and policy makers than the AI paradigm.