School of Engineering Department of Computer Science and Engineering 115 Interaction Design for Human-AI Collaboration Supervisor: MA, Xiaojuan / CSE Student: YAO, Chongchong / DSCT Course: UROP2100, Fall In recent years, many Go players have started to learn from artificial intelligence (AI) and imitate their moves. However, since many AIs are too powerful for humans to understand, it is a good idea to create AI agents with the same skill levels as the players and let the AIs collaborate with them. We pursue this goal by building a system consisting of three AIs: one that provides move suggestions, one that predicts the player’s moves prior to the player’s decision, and the other is a superhuman AI used as the ground truth to determine whether the player should make their own moves or listen to the AI agent. Interaction Design for Human-AI Collaboration Supervisor: MA, Xiaojuan / CSE Student: YUNG, Ka Shing / COMP Course: UROP1100, Spring Emotional support plays a crucial role today. Losing a close friend or family member can be an emotionally challenging experience, often requiring support and comfort. A virtual assistant can serve as an emotional companion, acting as a digital human, facilitating emotional communication and support through natural language generation, sentiment analysis, and Text-to-Speech (TTS) technology. However, existing chatbot systems still have limitations in sentiment analysis and explainability. Therefore, we propose utilizing the BERT model for sentiment analysis, incorporating LIME technology for explainability, levering generative AI for text, and employing TTS technology to generate a voice with specific emotions, which becomes a customizable digital human. Interaction Design for Human-AI Collaboration Supervisor: MA, Xiaojuan / CSE Student: ZHANG, Lveyang / COMP Course: UROP1100, Spring This report outlines the progress of our project to create a more engaging chatbot using the advanced large language model, ChatGPT, as its core. By leveraging the API of ChatGPT, we have achieved an impressive conversational ability for our chatbot and have added speech recognition capabilities by integrating it with a speech recognition project. Additionally, we have established a speech library that stores various speech models to provide personalized outputs at the output end. We have completed the backend program of our chatbot and encountered some practical issues during its running. Finally, we summarized the direction of future optimization, based on the problems encountered. And hope to apply the completed product for more immersive oral practice for users.
RkJQdWJsaXNoZXIy NDk5Njg=