| Summary: | In the rapidly evolving landscape of e-commerce, customer service stands as a pivotal component of consumer satisfaction and business success. This project centers on developing a system to assess the effectiveness of Large Language Models (LLMs) in augmenting the intelligence of chatbots for e-commerce sellers. With the digital marketplace in Singapore experiencing significant growth, alongside an increase in consumer complaints regarding online transactions, there is an urgent need for more responsive and sophisticated customer service solutions. By tweaking and integrating LLMs, particularly OpenAI's GPT-3.5 Turbo, with a chat web application, this research aims to enhance chatbot effectiveness and user experience. The study employs a methodological approach involving the development of a basic web app interface using bubble.io, integration with multiple AI models via OpenRouter.ai, and advanced training with Pinecone.io to facilitate vector embeddings for improved chatbot responses. Results indicate that LLM-enhanced chatbots demonstrate great performance in problem-solving, understanding context and nuance, and multilingual capabilities, thereby offering significant potential to improve e-commerce chat experience by providing more accurate, contextually relevant, and personalized or customized user interactions. This project not only fills a gap in understanding the impact of pre-trained language models on chatbot performance but also proposes practical solution for e-commerce owners aiming to improve their customer interaction and satisfaction levels, ensuring Singapore remains at the forefront of e-commerce innovation and adaptability.
|
|---|