STTR Phase I: Colleague: An AI-Enhanced Assistant Empowering K-12 Teachers with High-Quality Math Instruction

The broader/commercial impact of this Small Business Technology Transfer (STTR) Phase I project will contribute to the development of the nation’s STEM workforce through enhancing K-12 students’ mathematical proficiency beginning with support for educators. The artificial intelligence (AI)-enhanced platform augments K-12 teachers’ capacity to develop high-quality and inclusive math instruction to meet diverse students’ learning needs. The innovation leverages rapid advancement of AI in transforming both the workforce and educational landscapes, providing educators with a companion to significantly improve students' math performance, critical thinking skills, and readiness for a future AI-enhanced workforce. Importantly, this platform will democratize access to high-quality educational resources, particularly benefiting teachers in under-resourced schools by alleviating their workload stress and fostering a community of shared knowledge and practices. The adoption of this platform offers significant commercial opportunities within a $15bn market for educational technology in K12 education and offers a model of industry-university partnership in the development of educational technology, providing a robust and research-based solution to enhance scientific innovation and practical, classroom-based applications of AI while keeping humans in the loop through participatory co-design with educators. The project not only aligns with the national interest by promoting scientific progress and educational excellence but also holds substantial promise for economic returns. This Small Business Technology Transfer (STTR) Phase I project aims to tackle the pressing challenges in K-12 math education in terms of widened achievement gaps among student demographics, by developing education-specific AI technology. Phase I research and development will integrate new AI models which assist teachers in their ability to retrieve or generate lesson materials, catered to teachers’ instructional approaches and their students’ learning needs, and meeting research-based math instructional quality criteria. Personalized instruction, including formative assessment, auto-scoring and generation of diagnostic reports, targeted materials for enhanced student mastery or remediation as well as student feedback will be infused into the platform through iterative, participatory co-design student with 30 math educators and A/B testing with thousands of educators. Utilizing retrieval and augmentation models, nudging algorithms, and domain specific generative AI models that work alongside teachers to inspire creativity, agency, and growth that works alongside teachers, acting as a trusted companion, to prompt teachers to refine parts of the drafty lesson plan. This platform aims to shift the nature of AI-powered instructional technology with research-based practices tailored to domains and provided to educators in real time to transform how educators develop lesson materials and amplifies their ability to provide learners with more effective, personalized, and engaging instruction. 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.