Comparative Analysis of Artificial Intelligence Chatbot Performance

Abstract

Currently, Artificial Intelligence has penetrated almost every application and site on the Internet. This has had an impact on the world of education, business, etc. In addition to making it easier for humans to carry out their activities, this can also have negative effects if the response from the AI ​​chatbot is inaccurate or even misleading. In this article, the author has conducted experiments on several free versions of popular chatbots, namely ChatGPT, Gemini (Bard), AI Chat and Aichatting. AI chatbots are very helpful in the field of general knowledge, finding solutions to problems and creating creations, such as fiction, or guiding the creation of scientific reports. Users must understand and be observant in using this chatbot because the response is very dependent on the input given. Questions that contain errors will get different responses. All responses given by the AI ​​chatbot are suggestions, not something that must be followed absolutely. This is because the user is responsible for the consequences, not AI. AI has ethics in providing responses so that it will not produce negative answers or direct users to things that violate the rules. The factors that are the main measurements are speed, accuracy, completeness, understanding of questions, response variations, ability to handle input errors and ability to provide solutions. The four chatbots tested each have their own strengths and weaknesses, but there is one that has the most features and variations in its responses, namely Gemini (Bard).