Unraveling Indonesian Public R&D Institutions’ Perspectives on Chatgpt: An Analytical Approach To Decoding Open-Ended Surveys Through Topic Modeling

Abstract

<p>The rapid development of artificial intelligence (AI) technologies like ChatGPT has made global management easier. Since 2020, AI has been widely utilized for all purposes, notably due to its ease in access for scientific production. This study aims to analyze open-ended questions and responses from web-based survey in Indonesian public research and development (R&D) institutions using topic modeling. A total of 205 data points were obtained through web-based surveys conducted among researchers in Indonesian public R&D institutions. To learn their thoughts on the use of ChatGPT (Chat Generative Pre-Trained Transformer), two AI language topic modeling, Latent Dirichlet Allocation (LDA) and Principal Component Analysis (PCA), were employed to detect survey topic structures and show the results. Three theme groups represent the institution's research cohort. Based on generational disparities in birth year and functional position level, this study selected the seven most popular topics from three themes of researchers' opinions on ChatGPT. Researchers of certain age generations and functional position levels focused on new AI technologies, efficiency, and production gains, while others valued methodological innovation, ethics, and scientific integrity. When formulating a strategy for incorporating AI into the public R&D institution's future research agendas, it is imperative to include diverse perspectives.</p>