Understanding Teachers’ Use of Generative AI in Math and Science Instruction

• In Event: Innovating K–12 Teaching and Learning with Generative AI: Insights From IES-funded National R&D Centers

Abstract

This study investigated the role of generative AI (GenAI) in K–12 math and science teachers’ work using the RAND American Educator Panel, a nationally representative panel of U.S. public school teachers. Respondents included teachers who indicated that their primary subject area was mathematics, science or elementary education. Respondents who indicated that they taught elementary education were included in the survey if they also indicated a secondary subject area of mathematics or science. Nearly 1,000 teachers’ survey responses captured how educators incorporate GenAI into planning, instruction, assessment, professional learning and administrative tasks. GenAI was defined for respondents as digital tools that generate original content or automate tasks using large data patterns.

Our research contributes to prior surveys to fill out a picture of GenAI adoption in K-12 education (Dilberti et al., 2024; Gallup, Inc., 2025). These prior surveys demonstrate how science teachers use GenAI more than mathematics teachers. Our study deepens the inquiry by exploring the different GenAI use purposes and adoption curves (Rogers, 2010) of science and mathematics teachers. We explored:
1) To what extent do mathematics and science teachers use GenAI for various instructional tasks?
2) What differences exist between how mathematics and science teachers use GenAI for instructional tasks?
3) What are the key barriers to GenAI adoption for mathematics and science teachers?

Preliminary analyses reveal significant differences in adoption and use patterns between mathematics and science teachers. A greater proportion of science teachers reported using GenAI tools regularly for instructional tasks (See Figure 6). Figure 7 shows that the most common use is for lesson planning with 38% of teachers overall, followed by creating student assignments at 31% (note however that only 7% of teachers report using GenAI tools to assess student work). Interestingly, the usage gap between math and science teachers closed when it came to automating or assisting with administrative tasks and professional learning.

In the overall sample, less than half of teachers reported using GenAI tools. Teachers reported barriers to adoption including time required to learn and insufficient training opportunities (See Figure 8). Notably, two key potential barriers - perceived low quality of GenAI output and restrictive school district policies were only reported by 10% of teachers in the survey. While restrictive policies were not an issue, nearly half of teachers reported that unclear district guidelines or policies were a barrier to the adoption of GenAI tools, along with more than half of teachers reporting insufficient training opportunities and time required to learn.

This poster will present descriptive findings and comparative analysis of GenAI adoption by mathematics and science teachers. As technology providers, school district officials, teachers, and policymakers navigate GenAI’s rapid development and impact on learning, these findings present timely evidence about how and to what extent math and science teachers are adopting GenAI tools.

Authors

• Shawon Sarkar, University of Washington

• Lief Esbenshade, University of Washington

• Drew Nucci, WestEd

• Sarah Nielsen, WestEd

• Ann R. Edwards, WestEd

• Joshua Rosenberg, University of Tennessee

• Alex Liu, University of Washington

• Zewei Tian, University of Washington

• Zachary Zhang, Colleague.ai

• Kevin He, Colleague.ai

• Min Sun, University of Washington