AI Chatbot that Scrapes Website Information for
New Answers and Creates Articles

During the course of this study, we were curious about what some potential factors may be when it comes to an individual rating the performance of an AI chatbot. This curiosity stems from the idea of implementing a chatbot to help users to quickly get information or to navigate a website they are on. The sponsor of the project, Mr. Dave Burnett, also wanted to test out responses that were article length to see if there was an increase in descriptiveness as well. When planning this study, I developed a fully-functioning chatbot that utilizes OpenAI’s assistant APIs, the Scrapy library, the Tkinter library, and the WordPress library to develop code for the graphical user interface of the software along with the functions that generate the responses and format the responses based on the content scraped from the sponsor’s website AOKmarketing.com. For the study, we asked participants to test out the chatbot with a variety of prompts. For each of these prompts, the participants would rate the chatbots responses, which would then lead to the exit-survey. The exit-survey contains questions used to rate the chatbot along with demographic questions to test out a variety of demographic factors to see if they have an impact on the participants rating on the chatbot. After conducting the survey with a group of 40 people, we found that none of the demographic factors had any significant impact on the participants rating of the chatbot. The only factor that was close to having a significant impact was a person's familiarity with generative AI. At the end of the study, the importance of having a diverse group of participants, having quantitative and qualitative data, and making survey instructions as clear as possible. We were limited to only computer science students and did not have a big budget to improve the quality of the software. In future studies, we would want to have a vast demographic audience range and improve the software with up-to-date technologies.