A prototype of a chatbot for evaluating and refining student startup ideas using a large language model

Abstract

Assessing the soundness of business models is a critical skill for aspiring entrepreneurs and is an essential part of entrepreneurship education. However, evaluating business models can be time-consuming, costly, and subjective. This study describes the design and the prototype of a chatbot as a conversational intelligent tutoring system that assesses and gives feedback on business model soundness using natural language processing techniques and GPT-3.5, a large language model (LLM) trained by OpenAI, to help student co-founders learn and refine their startup ideas. Our method involves indexing articles and rubrics for evaluating technology startup pitches by extracting word embeddings via the OpenAI API. The chatbot accepts descriptions of startup businesses from student co-founders through a Telegram chatbot, and these are formatted as prompts and then fed into GPT-3.5. The responses are formulated by GPT-3 based on another set of prompts instructing the bot to give feedback from three virtual panelists: 1) a harsh judge, 2) a neutral expert, and 3) an optimistic investor.

Advancing Educational Accessibility: The LangChain LLM Chatbot's Impact on Multimedia Syllabus-Based Learning

“…This study places a central emphasis on LangChain, strategically designed to streamline the creation of customized AI applications utilizing LLMs. LangChain has received recognition within the AI community for its effectiveness in seamlessly interfacing with various data sources and applications [2]. The research provides a thorough examination of LangChain's fundamental features, shedding light on its components and chains.…”