AI-Ed Fusion: Symposium on STEM Education in the Era of AI

Dr. Wayne E. Wright, Associate Dean for Research, Graduate Programs and Faculty Development; and Professor and Barbara I. Cook Chair of Literacy and Language, College of Education, Purdue University

Dr. Kristen DiCerbo, Chief Learning Officer, Khan Academy

The rush is on to incorporate AI in STEM education. However, before we move blindly into the future, looking at the legacy of research in the field will help us build applications with a better chance of improving outcomes. This webinar will give a behind the scenes look at how AI features were built into Khan Academy, including how learning research is embedded into the design. You will hear what is being learned from the rollout of these features to schools and districts and get a chance to consider the opportunities and potential pitfalls of AI in STEM education.

An Approach to Developing a Course on AI by Leveraging Conjecture Mapping

Amogh Sirnoorkar, Ross-Lynn Postdoctoral Scholar, Department of Curriculum and Instruction, College of Education, Purdue University

Advancements in artificial intelligence (AI) have promised huge potential to transform the landscape of STEM education including curriculum, instruction and assessment. Given the significance, there is a need for developing courses that assist students in effectively leveraging AI platforms for their own learning. However, developing such a course presents several challenges including rapidly emerging technologies and unprecedented pace of novel research. In this talk, I present an approach to developing such a course through the lens of a design-based education framework called “Conjecture Mapping” that navigates through these challenges. I contextualize the discussions in the recently designed course titled “Artificial Intelligence in STEM Education” at Purdue. Lessons learnt in designing this course and future course planned for the upcoming academic year are discussed.

AI Literacy Competencies for K-12 Teachers: A Systematic Literature Review

Tirtha Karki, PhD student, Department of Curriculum and Instruction, College of Education, Purdue University

Collaborators: Deepti Tagare and Wonjin Yu

As AI technology becomes prevalent, people’s attention has been drawn to AI more and more, especially in the field of education. Accordingly, research has begun to discuss the need for including AI education in K-12 settings to prepare students from a young age for the changing nature of the workforce. This has led to defining ‘AI literacy’ as the ability to learn and effectively use the basic principles of AI. However, since teachers teach AI to their students, teachers first need professional development to develop content, pedagogical and technical knowledge about AI and be prepared to provide opportunities for learners to achieve proficiency in AI literacy. Therefore, there is a need to analyze, discuss, and consolidate teachers’ AI literacy competencies in terms of the knowledge, skills, and dispositions needed to teach AI literacy. This systematic literature review uses the “CoLeaf” competency framework to identify the knowledge, skills, and dispositions that teachers need to integrate AI literacy in K-12 education consolidated from studies on teachers’ AI literacy. This talk discusses preliminary analysis of 817 peer reviewed papers published during 2014-2024.

Facilitating Analogical Reasoning in Physics Using Generative-AI

Dr. Ian Arnold, Assistant Professor, Department of Physics and Astronomy, College of Science, Purdue University

Collaborators: Sanjay Rebello and Amogh Sirnoorkar

Making sense of conceptual ideas through analogies is a common practice in physics. In this talk, I present preliminary results on the use of Generative-Artificial Intelligence (AI) in facilitating students’ understanding of physics concepts through analogical reasoning. The study involves students articulating their understanding of Coulomb’s physics law in three ways: (i) in their own words, (ii) as though explaining to a second grader, and (iii) using scenarios of their choice. Students then compare their explanations with responses generated by ChatGPT for the same tasks. In addition, students also make qualitative observations between the two sets of descriptions. I discuss the results in light of the affordances and constraints of using AI platforms in teaching and learning of physics.

Analyzing Contemporary Policies Surrounding the Use of AI in U.S. Higher Education Institutions

D’Angelo Peters, PhD student, Department of Curriculum and Instruction, College of Education, Purdue University

Collaborator: Ravishankar Chatta Subramanium

Artificial Intelligence (AI) has been making inroads into all aspects of human life including education. The release of AI tools such as ChatGPT and Bard has hastened the need for universities to take a detailed look into issues related to the use of AI. In this article we compare and contrast the AI policies formulated by Midwestern higher education institutions and universities. Given that the scope of the study is vast, we restrict ourselves to three aspects components derived from Chan’s (2023) AI Ecological framework: use, deployment, and ethical considerations regarding the use of AI. We note that a significant number of universities are yet to form an AI policy. With institutions beginning to publish their policies, we note that the policies may be treated as living documents which will grow as AI is better understood. Our study provides insights into how universities are approaching the issue, which in turn would increase AI awareness for all stakeholders.

Development, Validation, and Analysis of Three-Dimensional Learning Assessments in Physics using Generative-AI

Dr. Sanjay Rebello, Professor, Department of Curriculum and Instruction, College of Education; and Department of Physics and Astronomy, College of Science, Purdue University

Collaborators: Amogh Sirnoorkar and Sai Munikoti

Recent reports in higher education have called for shifting the focus on contemporary science learning towards promoting authentic knowledge-building practices. To facilitate this objective, frameworks such as “Three-Dimensional Learning” are advocated. This framework characterizes science learning along three “dimensions” namely (i) disciplinary core ideas – ideas central to understanding of a discipline, (ii) cross-cutting concepts – concepts that span across multiple disciplines, and scientific practices – disciplinary practices that are key to generating new knowledge in science. However, developing these assessments present several challenges including contextualizing them in real-world scenarios. We address these challenges by leveraging Generative-AI through a prompt template customizable to facilitate assessment development based on the instructors’ choice of content areas, scientific practices, core ideas, and cross-cutting concepts. Insights from piloting the generated assessments in introductory courses are discussed.

Enhancing STEM Education Through AI-XR Integration: A Systematic Literature Review of Trends, Applications, and Challenges

Koki Motoi, Graduate Student, Department of Curriculum and Instruction, College of Education, Purdue University

Collaborators: Belle Li, Cansu Coskun, and Anthony Chuka Ilobinso

Recent advancements, such as generative Artificial Intelligence (AI) and advanced Extended Reality (XR) platforms, are pushing the boundaries of the combination of AI-XR. These developments hold the promise of transforming education, making it more accessible, efficient, immersive, collaborative, self-directed, and tailored to individual learning paths. Despite the surge in interest, there is a shortage of extensive reviews on this integration. This systematic literature review aims to (1) explore the current state of AI-XR technology integration in STEM education, focusing on specific STEM subjects, targeted skills, and variations across different STEM disciplines, (2) investigate how AI and XR technologies are being integrated in STEM education, examining the specific types of AI and XR technologies utilized and their contributions to the learning process, (3) address the challenges and barriers to integrating AI-XR technologies in STEM education, and (4) identify gaps in the research that warrant further investigation.

Exploring the Values Invoked by LLMs in Response to Varying Prompts Surrounding Ethical Scenarios

Dr. Justin Hess, School of Engineering Education, College of Engineering, Purdue University

Collaborators: Tyler Garcia and Amogh Sirnoorkar

The widespread adoption of Generative-Artificial Intelligence has attracted significant traction amongst educators’ community. Given students’ wide usage of AI on a range of domains including ethical scenarios, there have been efforts to explore the “moral dimension” of AI-generated responses. In this exploratory work, I present the frequency and spectrum of values evidenced by AI platforms to typical ethical scenarios experienced in STEM education settings. The results are contrasted to scientists’ reasoning on similar scenarios reported in the literature.

Content Analysis of Pedagogical Relationships Involving Generative-AI Supported Teaching and Learning Practices in Physics Education

Marla Grover, PhD student, Department of Physics and Astronomy, College of Science, Purdue University

Collaborator: Zeynep Akdemir

This work-in-progress study reviews literature on the use of Artificial Intelligence in Education (AIEd) within the context of physics education, to understand its impact on the interplay between curriculum, teaching, and learning. Recognizing the myriad possibilities and challenges of future human-AI interactions, we seek to review existing research positions of AIEd as a mediator that changes pedagogical elements in physics education. Our review encompasses relevant research articles, conference proceedings, and book chapters from the last decade. We delve into critical interactions such as students’ engagement with content, discussions with teachers and peers, hands-on work with laboratory equipment, and moments of insight or discovery. From the educators’ perspective, we explore how AI facilitates deeper understanding of content, preparation of instructional activities and materials, provision of student feedback, and assessment of student understanding. Our analysis aims to highlight AI’s transformative potential in physics education, fostering future innovations that bridge technology and pedagogy. By elucidating the triangular relationship between curriculum, teaching, and learning, we contribute to the dialogue on optimizing AI’s role in educational settings, ultimately seeking to improve teaching practices and student learning outcomes in physics education.

Developing Coupled, Multiple-Response Assessments by Leveraging Generative-AI in Physics

Ravishankar Chatta Subramanium, PhD student, Department of Physics and Astronomy, College of Science, Purdue University

Collaborators: Amogh Sirnoorkar and Sanjay Rebello

Assessments are central to academic practice, particularly in Science, Technology, Engineering, and Mathematics (STEM) courses. In this talk, I present an approach to developing an assessment format called “Coupled Multiple-Response (CMR)”. This format entails multiple-choice and multiple-response formats paired together to facilitate students to both committing to a claim along with selecting options that align with their reasoning. In addition to facilitating streamlined scoring, this format captures subtle nuances beyond correctness of solutions. The assessment is developed by augmenting student data from an ongoing physics course along with Generative-AI. An approach to leveraging AI in developing this assessment is discussed.

Dr. William R. Watson, Director of the Purdue Center for Serious Games and Learning in Virtual Environments; and Professor of Learning Design and Technology, College of Education, Purdue University