New Perspectives in Science Education

As generative artificial intelligence (AI) continues to integrate into the educational landscape, its potential to transform teaching and learning practices becomes increasingly evident. In the context of chemistry education, we are developing an AI-based assistant specifically tailored to the needs of classroom instruction. This system leverages fine-tuned models based on OpenAI’s large language model (LLM) ChatGPT. The focus is on harnessing AI to address key challenges in education, including personalizing learning experiences, differentiating instructional content, and automating grading and feedback processes.

Our research involved collecting training data on classical topics in chemistry education, such as chemical bonding, atomic structure, and chemical reaction. Additionally, we gathered insights from educators to identify their specific needs and expectations regarding AI tools in teaching. Based on these inputs, the AI assistant aims to enhance learning efficiency while also serving as a supportive tool to reduce the workload of educators.

In our presentation, we provide an overview of the development and implementation of this AI system. We demonstrate how fine-tuning LLMs can enhance the quality and relevance of AI-driven educational support. Beyond the technical development, we explore practical applications within chemistry classrooms, showcasing how AI can enable personalized and effective learning experiences while complementing the role of human educators.

Furthermore, we discuss the broader implications of integrating AI into education, including challenges related to teacher and student acceptance, ethical considerations, and data privacy concerns.