Visualising AI : a game presenting neural networks

Abstract

Artificial Intelligence (AI) is becoming a more integral part of daily life, yet many people still have very little insight on it. Machine Learning makes up a large portion of AI technologies and often makes use of black box models which make it even harder for users to understand and trust the process. Education on such important topics is therefore increasing in importance. Visualisation and interactivity through digital games for learning can provide a great foundation for tackling this, and some projects have already looked into creating educational games to provide users with a stronger idea of how certain AI models work. This FYP expands on this through the research and development of a visual game like application to educate and present the topic of Neural Networks. The aforementioned application involves the implementation of Neural Network that can interface with an endless runner game, and a user interface designed with explainability in mind that allows users to interact and experiment with the Neural Network. By making use of this, users can collect data, train and optimise the Neural Network to have it play the endless runner game, thereby learning more about the structure and setups of Neural Networks through game play and experimentation, rather than from a textbook. Evaluation is carried out through a user study, where participants are presented with the game, as well as accompanying pre and post surveys. These surveys have been adopted and adapted from previous studies and aim to gauge how effective the application is in illustrating these concepts. The results indicate that such an application is quite effective to help people understand more complex concepts, with the educational game scoring well on a standardised scale, and the average understanding and interest in Neural Network concepts increasing significantly amongst participants.