Artificial Intelligence Generative Tools and Conceptual Knowledge in Problem Solving in Chemistry

Abstract

In recent years, artificial intelligence (AI) has emerged as a valuable resource for teaching and learning, and it has also shown promise as a tool to help solve problems. A tool that has gained attention in education is ChatGPT, which supports teaching and learning through AI. This research investigates the difficulties faced by ChatGPT in comprehending and responding to chemistry problems pertaining to the topic of Introduction to Material Science. By employing the theoretical framework proposed by Holme et al., encompassing categories such as transfer, depth, predict/explain, problem solving, and translate, we evaluate ChatGPT’s conceptual understanding difficulties. We presented ChatGPT with a set of thirty chemistry problems within the Introduction to Material Science domain and tasked it with generating solutions. Our findings indicated that ChatGPT encountered significant conceptual knowledge difficulties across various categories, with a notable emphasis on representations and depth, where difficulties in representations hindered effective knowledge transfer.

Kajian Literatur Pemanfaatan Teknologi Artificial Intelligence untuk Meningkatkan Keterampilan Abad 21 Siswa dalam Pembelajaran Kimia

“…Akan tetapi, penggunaan Artificial Intelligence memiliki kekurangan lain yaitu pada website ChatGPT masih belum bisa menyelesaikan semua pertanyaan kimia. ChatGPT kesulitan memberikan solusi yang berkaitan dengan struktur kimia yang kompleks (Daher et al, 2023). Sementara itu, media game juga dapat dijadikan solusi untuk membantu siswa memahami dan meningkatkan pengetahuan tentang kimia.…”

Does ChatGPT Play a Double-Edged Sword Role in the Field of Higher Education? An In-Depth Exploration of the Factors Affecting Student Performance

“…They can learn from data through classification, prediction, and generation, subsequently assisting students in completing tasks, making judgments, and tailoring individualized learning paths based on students' capabilities [39,40]. During this process, GAI educational tools can evaluate the effectiveness of students' learning, such as the creativity demonstrated in artistic activities [41] and the depth of understanding of fundamental chemical concepts [42]. They can help students in identifying and rectifying gaps in their knowledge, laying a robust foundation for subsequent in-depth studies, thus promoting the achievement of sustainable educational development objectives [38].…”

The Future of ChatGPT in Medicinal Chemistry: Harnessing AI for Accelerated Drug Discovery

“…[2] These platforms support the capabilities of machine learning, predictive modeling, and virtual screening to rapidly analyze vast datasets and predict the behavior of potential drug candidates. [3] For instance, In silico Medicine is using AI to create an entirely new AI-driven drug discovery, predict their interactions with biological targets, and optimize properties for efficacy and safety. [4] Schrödinger's suite of AI-driven tools empowers researchers to simulate molecular interactions, predict binding affinities, and guide lead optimization, significantly accelerating the iterative process of drug design.…”

“…Cutting‐edge AI‐powered platforms such as Chemistry42, In silico Medicine, Schrodinger, and Cyclica, are also reshaping the drug discovery process from molecule identification to optimization [2] . These platforms support the capabilities of machine learning, predictive modeling, and virtual screening to rapidly analyze vast datasets and predict the behavior of potential drug candidates [3] . For instance, In silico Medicine is using AI to create an entirely new AI‐driven drug discovery, predict their interactions with biological targets, and optimize properties for efficacy and safety [4] .…”