Enhancing AI\nResponses in Chemistry: Integrating Text\nGeneration, Image Creation, and Image Interpretation through Different\nLevels of Prompts

Generative Artificial Intelligence technologies can potentially\ntransform education, benefiting teachers and students. This study\nevaluated various GAIs, including ChatGPT 3.5, ChatGPT 4.0, Google\nBard, Bing Chat, Adobe Firefly, Leonardo.AI, and DALL-E, focusing\non textual and imagery content. Utilizing initial, intermediate, and\nadvanced prompts, we aim to simulate GAI responses tailored to users\nwith varying levels of knowledge. We aim to investigate the possibilities\nof integrating content from Chemistry Teaching. The systems presented\nresponses appropriate to the scientific consensus for textual generation,\nbut they revealed alternative chemical content conceptions. In terms\nof the interpretation of chemical system representations, only ChatGPT\n4.0 accurately identified the content in all of the images. In terms\nof image production, even with more advanced prompts and subprompts,\nGenerative Artificial Intelligence still presents difficulties in\ncontent production. The use of prompts involving the Python language\npromoted an improvement in the images produced. In general, we can\nconsider content production as support for chemistry teaching, but\nonly with more advanced prompts do the answers tend to present fewer\nerrors. The importance of previously understanding chemistry concepts\nand systems’ functioning is noted.