Measuring the Prowess of Artificial Intelligence in Classifying Veterinary Books

Abstract

This study aims to measure the accuracy of classification capabilities of Artificial Intelligence (AI) powered chatbots and determine the extent of deviation of AI-generated class numbers from the standard class numbers in the discipline of Veterinary Science. A total of 50 most used books in Veterinary Science at Guru Angad Dev Veterinary and Animal Sciences University (GADVASU), Ludhiana, were selected for this study. These book titles were classified using the 22^nd edition of the Dewey Decimal Classification (DDC) scheme by Library and Information Science Professionals (s) (LISPs). Then, class numbers of these titles were searched in three prominent AI chatbots, viz. ChatGPT, Gemini and Copilot, using a standard search string. The numbers so obtained were tabulated, and comparisons were made to find the accuracy of AI-generated class numbers and the extent of deviation from class numbers assigned by LISPs. The results of the study revealed that AI chatbots had very low rates of accuracy in classifying books of veterinary science. A thorough examination of the results exposed strong deviations in the AI-generated class numbers from the class numbers assigned by LISPs. This study delineates the challenges uncovered while classifying veterinary science books using AI Chatbots and concludes that human involvement in the field of classification cannot be overlooked. Therefore, deep insights into the usage of AI in classifying documents of other disciplines should be explored before establishing any kind of generalisation.