Leveraging Generative Artificial Intelligence for Analyzing Handwritten Chemistry Notes: A Chatbot Support for Students’ Queries
摘要
The use of Generative Artificial Intelligence systems has been reported in different fields of investigation. In the educational area, some studies have focused on the possibilities of this use. Considering the importance of understanding the potential and limitations, this study aimed to investigate the content generated by ChatGPT 4.0 in interpreting handwritten notes, simulating the potential use of this tool as a chatbox for queries. Accordingly, our research questions are: What is the level of correctness of the content generated by ChatGPT, in its paid version, for higher-level organic chemistry questions? Can this content generated by ChatGPT be used as feedback in teaching situations? Based on images produced for higher-level organic chemistry content were analyzed through 24 prompts, the answers to which had their content analyzed. The analysis was carried out using the content analysis technique, which considered exclusive and pre-established categories. The results indicate that ChatGPT 4.0 can properly recognize and interpret different symbolic aspects of chemical language. Some errors are related to common symbols, such as arrows and flat balls, which have a specific meaning in chemistry. ChatGPT presents more errors than image interpretation for additional and more complex questions, showing the need for training to improve the chatbox. However, due to the number of correct answers, we see potential for use as an additional tool for teaching Chemistry.