Leveraging Generative AI for Automatic Scoring in Chemistry Education: A Web Based Approach to Assessing Conceptual Understanding of Colligative Properties

Authors

  • Sri YAMTINAH Chemistry Education, Universitas Sebelas Maret Author
  • Dimas Gilang RAMADHANI Chemistry Education, Universitas Negeri Semarang Author
  • Antuni WIYARSI Chemistry Education, Universitas Negeri Yogyakarta Author
  • Hayuni Retno WIDARTI Chemistry Education, Universitas Negeri Malang Author
  • Ari Syahidul SHIDIQ Chemistry Education, Universitas Sebelas Maret Author

DOI:

https://doi.org/10.58459/icce.2024.4983

Abstract

The integration of artificial intelligence (AI) into educational assessment offers promising advancements in automating the grading process, particularly in complex subjects like chemistry. This study focuses on implementing the Gemini 1.5 AI model to evaluate student responses in a web-based chemistry assessment. The study aimed to assess the effectiveness and accuracy of Gemini 1.5 in grading questions related to stoichiometry, a fundamental concept in chemistry. The assessment involved 320 students who answered five questions one conceptual and four computational—focused on calculating molar quantities and applying related formulas. The AI system was utilised to evaluate the responses, providing scores based on criteria such as the correct application of formulas, calculation accuracy, and the proper use of scientific units. The study's findings indicate that Gemini 1.5 demonstrated high accuracy, with precision and recall metrics consistently ranging from 0.87 to 0.93 across the different questions. These results suggest that the AI system effectively delivered consistent and objective grading, minimising errors such as false positives and negatives. The AI's ability to provide immediate and detailed feedback highlights its potential to enhance learning by reinforcing key concepts and addressing areas where students may struggle. The conclusion drawn from this study is that integrating Gemini 1.5 into the educational assessment process improves grading efficiency and supports personalised learning by offering tailored feedback. This integration has significant implications for reducing the workload on educators while ensuring fair and accurate assessments, ultimately contributing to a more effective educational experience for students.

Downloads

Download data is not yet available.

Downloads

Published

2024-11-25

How to Cite

Leveraging Generative AI for Automatic Scoring in Chemistry Education: A Web Based Approach to Assessing Conceptual Understanding of Colligative Properties. (2024). International Conference on Computers in Education. https://doi.org/10.58459/icce.2024.4983