AI-Based Identifier Tools in Science Education: A Thematic Review of Accuracy, Pedagogical Potential, and Perception Widi Purwianingsih (a*), Pipih Nurhayati (b**), Ari Widodo (b), Riandi (b), Achmad Samsudin (c)
(a) Biology Education Program
Universitas Pendidikan Indonesia, Bandung, Indonesia
*widipurwianingsih[at]upi.edu
(b) Science Education Program
Universitas Pendidikan Indonesia, Bandung, Indonesia
**pipih.nurhayati[at]upi.edu
(c) Physics Education Program, Bandung, Indonesia
Abstract
The development of artificial intelligence (AI)-based applications, especially in the form of AI identifier tools such as plant recognition applications and image-based classification systems, has opened up new opportunities in science learning on environmental issues. This study aims to analyze articles related to the use of AI-based identification tools, ranging from plant image classification, water quality detection, to strengthening AI literacy among students and prospective teachers. The analysis was conducted to study the accuracy, pedagogical potential and user perception of students, teachers and prospective teachers. To obtain relevant articles, the SPIDER inclusion criteria and PRISMA framework from the SCOPUS database from 2020 to 2025 were used. In the end, this study analyzed 15 of the 121 articles obtained. The analyzed research shows that applications such as PlantNet, Google Lens, and deep learning-based tools are able to provide high identification accuracy, including the ability to determine the type of plant species. However, the means and facilities to access these applications are still limited. Several studies have also highlighted the pedagogical potential of AI identifiers, particularly in encouraging interest in STEM learning, fostering environmental awareness, and developing digital literacy among students and teachers. In addition, teachers^ and prospective teachers^ perceptions of the use of AI-based identification applications indicate an opportunity for integration into authentic and contextual learning-based curricula that have become a global reference for science learning. This thematic review based on article analysis is expected to be a reference in developing research related to AI Identifier tools in science education.
