Implications of Artificial Intelligence for Astronomy Education
13:30 - 13:45
Co-author: Vive Kumar
The topic of artificial intelligence (AI) has been in the news a lot recently, in part due to potential misuse by students. Astronomy is arguably a good subject for use of AI due to its varied content in which knowledge of specialized terminology is important. For example, multiple choice tests are arguably more effective in testing astronomy learning than in, for example, physics, with its more limited introductory number of concepts, but usually assessed through problem-solving competence. With this in mind we have conducted some tests using the online ChatGPT service, which allows entry of topics, with directives, to generate text output. The directives may include such statements as “generate a multiple choice test” and “in 100 words explain”, followed by a topic. Queries were also attempted at higher levels including expert knowledge expressed with mathematics. It is also possible to conduct a “Turing test” comparing human expert answers with those generated by ChatGPT. The positive outcomes of these tests frame possible help that ChatGPT can give to instructors. Generation of multiple choice astronomy tests gave results comparable to those from test banks, needing some slight correction before they could be considered for use with students. Short explanations generally were clear and could be used in making course materials. Negative outcomes included that students could likely generate answers for assignments without actually doing much work, or attaining understanding through study as is the intent. Progress in AI likely needs to be factored into astronomy education in both senses.