Assessment of learning using ChatGPT: lessons for teaching chemistry

Abstract

This work is the result of research that aimed to investigate, from the perspective of undergraduate chemistry students, the contributions of ChatGPT in the development of a question to assess the learning of a chemical content covered in high school. Learning assessment is one of the teaching activities, and since it can be developed with the aid of generative artificial intelligence, it is important that undergraduate students reflect on its limitations. To achieve this objective, we developed an assessment question within the scope of the subject Assessment of the Teaching and Learning Process in Chemistry, which is present in the curriculum of undergraduate chemistry programs at a federal university located in Minas Gerais, aligning the theme of assessment with the use of generative artificial intelligence. From the responses of the five undergraduate chemistry students enrolled in the subject, we constructed the data and interpreted it through content analysis. We found that the students identified that ChatGPT provided answers that need to be reviewed and corrected. This indicates that future teachers will not use the information obtained without verifying its accuracy, which suggests a learning process for teaching. Given the results obtained, we emphasize the importance of conducting further research, as there is still a gap in knowledge production regarding teacher training in assessment practices involving generative artificial intelligence.