Ever since the all-knowing AI demonstrated its capability to comprehensively respond to user inquiries at the end of 2022, businesses have been trying to utilize generative AI to build smart customer service and AI assistants as a means to reduce staff workload and increase work efficiency, while also boosting customer satisfaction.
Participating in the smart mobility: Gogoro division of the AI Wave: Taiwan Generative AI Applications Hackathon, held on May 18-19, 2024 by DIGITIMES, the team named "How Come Fu-Gui Left," comprising Fu-Gui Wang from Bachelor's Program of Artificial Intelligence Technology and Application, Feng Chia University, as well as Si-Ya Chen, Yan-Ting Guo and Cheng-Han Lin from Department of Information Engineering and Computer Science, Feng Chia University, created "Gogoro Smart Scooter Wizard," an AI assistant built with generative AI that gives instant and complete responses to all kinds of user inquiries about Gogoro. The team stood out from a group of strong competitors and won the honorable mention award, voted unanimously by all the judges.
The market has seen quite a few AI-based chatbots emerge in recent years. However, most of them only provide standardized answers to specific questions. In contrast, How Come Fu-Gui Left developed Gogoro Smart Scooter Wizard using the Anthropic Claude 3 Sonnet and Haiku models in Amazon Bedrock, a fully managed service that makes high-performing foundation models on Amazon Web Services (AWS), trained on massive data from Gogoro, allowing it to give organized and sensible responses. Users only need to input the Gogoro model to start asking any questions and get complete answers.
Despite its powerful capabilities, generative AI with large language models (LLMs) may occasionally give false or misleading information. When this happens, it does not help but burdens customer service agents.
To prevent this, How Come Fu-Gui Left implemented multiple measures in their AI assistant. First, it is restricted to providing answers only to Gogoro-related questions. Furthermore, to save the AI model from incorrectly identifying a Gogoro vehicle, the AI assistant searches the Gogoro knowledge base for information based only on the vehicle model entered by the user. It then sorts out relevant content and provides a user-friendly response. It will politely refuse to answer questions unrelated to Gogoro vehicles.
Gogoro Smart Scooter Wizard further uses generative AI to give answers with images. Not only will this help users better understand the response, but it also optimizes the user experience and raises user satisfaction, thereby boosting Gogoro's brand awareness and trustworthiness.
The team named "How Come Fu-Gui Left" won honorable mention for their AI assistant for Gogoro smart scooters