AI-Assisted Learning of Blended Math-Science Sensemaking Skills (AI4MSS)

Leonora Kaldaras

Blended Mathematical Sensemaking in Science (MSS) involves a deep understanding of quantitative relationships and scientific meaning of equations describing phenomena. MSS is a prerequisite for science mastery. While MSS have been described for specific disciplines, my recent work has formulated and tested a cognitive theory of MSS that applies across multiple scientific fields. I showed that the theoretical framework can guide the development of instructional sequences on MSS that helped most students in introductory Chemistry and Physics classrooms transition to higher proficiency levels. This scaffolding, paired with interactive PhET simulations that support this instructional approach, afforded essentially autonomous learning without the need for student-instructor interaction. This provides an attractive way to help learners develop MSS skills. Further, the framework has proven effective with a population of community college students from backgrounds historically marginalized in STEM across US because it provides for autonomous learning activities that meet the differing needs of individual learners. In the current study I am proposing to enhance the self-guided MSS learning activities I have shown to be effective for fostering MSS with generative AI (GAI) chatbot capable of providing immediate, tailored, and interactive MSS feedback to individual learners. I believe that this will further improve the effectiveness of self-guided learning activities in supporting MSS among students. This, in turn, has the potential to help improve learner success in introductory undergraduate STEM courses and improve the chances of STEM-related career outcomes among learners, ensuring success in STEM for all students.

About Leonora Kaldaras

Leonora (Lora) Kaldaras, PhD is an incoming Assistant Professor in Artificial Intelligence, Teaching and Learning at the University of Houston College of Education. Dr. Kaldaras previously held a dual position as a visiting scholar at Stanford University Graduate School of Education and PhET Interactive Simulations (University of Colorado Boulder). During this time, she worked with the founder PhET and a Physics Nobel Prize laureate Dr. Carl Wieman and the director of PhET Dr. Kathy Perkins. Her work focused on designing personalized learning experiences to foster development of complex cognitive processes (e.g., math-science sensemaking, knowledge transfer) through self-guided learning strategies using PhET. She will continue this work through the Spencer Postdoctoral Fellowship. Dr. Kaldaras is also co-PI on an NSF project (Award # 2200757) focused on developing LP-guided AI-driven formative feedback system for high school Physical Science. She studies ways to leverage AI to support learners in developing complex cognitive skills related to knowledge application and mathematical reasoning across STEM domains. She is pursuing several research directions related to this goal. First, she explores using AI to formulate cognition theories describing foundational cognitive processes of knowledge application and sensemaking and facilitating the use of these theories to guide the learning process. Second, she explores leveraging AI in supporting teachers and learners during the learning process to facilitate the development of knowledge application using learning progression-guided personalized, AI-supported scaffolds. Third, she studies bias and validity issues in AI-based scores on LP-aligned constructed response assessments measuring knowledge application in STEM.