This project aims to serve the national interest by designing and implementing an Artificial Intelligence (AI)-augmented formative assessment and feedback system. This system will help students develop source-based STEM arguments, such as STEM text summarization, or problem spaces, which are mental representations of a problem and of multiple paths to solving it. Project implementation will take place in large, undergraduate introductory physics courses at an urban university that serves diverse and historically underrepresented student groups. Persistent learner engagement in pre-classroom learning activities is critical to learner success in introductory STEM courses. Undergraduate students often need to develop a solid understanding of content or problem situations in self-paced online learning contexts to prepare for in-classroom active and collaborative learning. However, unsupervised pre-classroom learning can be an ongoing issue in a student-centered learning model. This problematic situation is particularly evident in large introductory-level STEM courses where traditional instructional techniques are less effective. The innovation of this project will include AI-generated adaptive scaffolding information and learning progress feedback with data visualization techniques to help students with conceptual learning and self-regulatory behaviors. The unique learning opportunities supported by an AI-scaffolded feedback system will significantly increase students' engagement levels in self-paced online pre-classroom learning. This, in turn, should help students acquire content knowledge and build a proper understanding of problems to prepare themselves for success with in-classroom interactive problem-solving activities.<br/><br/>Three phases will govern the work of this project. First, the project team will take a Participatory Research (PR) approach that emphasizes the direct engagement of faculty members who teach physics courses in designing and implementing new assignments. These faculty members will also co-construct research through a partnership with researchers to conduct a mixed-methods study of instructors and students in the courses. During this first phase the primary research goal is to identify topics and problems that utilize AI-scaffolded pre-classroom learning and investigate learner engagement and progression in the pre-class assignments. In the project's second phase evaluation studies will demonstrate whether knowledge development during pre-classroom learning can help students solve cognitively demanding tasks in classrooms and develop positive self-efficacy in STEM. The findings will also determine whether AI in education improves students' well-being inside and outside of classrooms, with a focus on students traditionally underrepresented in STEM education. Extensive data collected in the final phase will uncover the relationships among pre-classroom activities, in-classroom performance, self-efficacy, interest in physics, and student backgrounds, including gender, race, ethnicity, first-generation status, and English language learning. The sequence mining and cluster analysis are expected to reveal students' different hidden engagement states and group their engagement trajectories, explaining how cluster membership and trajectories vary across students' backgrounds. Consequently, this project will lay the groundwork for further research to develop an AI-scaffolded pre-classroom learning model that promotes most students' success in introductory physics courses. The NSF IUSE: EDU Program supports research and development projects to improve the effectiveness of STEM education for all students. Through its Engaged Student Learning track, the program supports the creation, exploration, and implementation of promising practices and tools.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.