Generative AI in STEM Teaching: The Current State and Emerging Trajectories

• In Event: Generative AI in STEM Teaching and Learning: Informing Transformative Uses Through Ethical Research and Development

Abstract

AI has been part of educational designs and research from potentially replacing teachers as 1-1 intelligent tutoring systems to supporting teacher decision-making through more sophisticated AI techniques, learning analytics, and education data mining. Recent machine learning techniques have ushered a new era in AI in STEM education. The arrival of LLMs and more “general-purpose” AI tools, directly usable by teachers without requiring them to hold specialized knowledge and thereby reducing any friction in their usability, has the potential to not only expand the uses and use cases of AI in the STEM classroom, but also potentially empower the teacher in ways that are still being explored and largely unknown.

Educators, learners, and researchers are grappling with unanticipated and rapidly changing AI that impacts teachers’ day-to-day K-12 classroom practices. In this paper, we highlight the current state of research and practice with respect to AI’s impacts on STEM teaching, and share emerging promising trajectories for research and practice around AI and STEM teaching. Through a survey of the nascent existing literature and current portfolio of research projects, we examine how AI is currently impacting the following key dimensions of STEM teaching: 1) Justice, Equity, and Inclusion, a central tenet in today’s STEM teaching for facilitating inclusive STEM learning environments that promote justice and equity (Author et al.; Watkins, 2021); 2) Curricular Decision-Making and Lesson Planning, a core component of creating successful STEM learning environments (Karpouzis et al., 2024; Wang, 2023); 3) Pedagogy and Instruction, focused on the time spent in the classroom working with students (Cooper, 2023; Demszky & Liu, 2023); 4) Assessing Student Work and Progress, to support ongoing student learning through measurement and feedback (Grover & Clarkson, 2024; Wang, et al., 2023); and 5) Teacher Learning and Leadership, so that preservice and inservice STEM teachers continue to grow and progress in their capacities in the service of students (Chiu, 2023; Langran et al., 2024). We also explore how evolving research can inform the design, development, and use of AI along these dimensions.

We then turn our attention to future trajectories, identifying critical themes in the ongoing research and design efforts around AI in STEM teaching. We assert that in order for education-focused AI to incorporate an ethic of justice, equity, inclusion, and caring for all learners, teachers and students’ communities representing a diverse range of identities, perspectives, and community memberships must have a voice in co-designing. We push for Generative AI-supported classroom assessment to use AI models that are trained in and reflect accurate models of student learning, situational awareness of classroom contexts and build on past research in learning progressions to promote successful STEM learning. Lastly, we outline a two-track approach to a research agenda that explores both the education of teachers in prompt engineering and other emerging skills to obtain more targeted and helpful results, as well as designing AI with more finely tuned and contextual training data, such as in the form of knowledge graphs (Tripathi, 2021).

Authors

• Jeremy F. Price, Indiana University - Indianapolis

• Shuchi Grover, Looking Glass Ventures