AI to the Test: Measuring ChatGPT's Objective Accuracy in the SATs in Comparison to Human Performance

In the information age, students today have access to numerous digital tools and sources that can help them learn lessons, write and check texts, and answer questions from a variety of subject areas. Tools such as paraphrasers, grammar checkers, and generative chatbots are all made possible today by artificial intelligence. While its rapid development has brought efficiency to various sectors, artificial intelligence also comes with its own set of drawbacks. ChatGPT, an advanced generative AI chatbot, has gained popularity as an educational aid for generating answers and explanations for different subject area questions, despite its risk of disseminating misinformation. While previous studies have assessed the performance of ChatGPT in various domains, such as law and medicine, there is a gap in research on the analysis of its efficacy in secondary school-level subjects. Therefore, this study aims to assess the accuracy of ChatGPT's responses and its tendency to provide incorrect information in linguistics and mathematics questions at the high school level. To achieve this, SAT questions are administered to ChatGPT, subjecting it to a standardized examination commonly used to evaluate a high school student's knowledge and aptitude. Later on, ChatGPT's performance will be compared to the global SAT test-taker population's performance, gender, and race.