How a New Math Pilot is Informing Our AI Learning Agenda

As educators across the K–12 landscape explore generative artificial intelligence (AI), many leaders are being pulled into decisions about tools before they’ve had the opportunity to determine how AI should fit into their core instructional strategy. Shifting this frame toward more strategic decision-making could yield greater promise for teaching and learning.

Math is one particular area of opportunity.

Overall, math outcomes, professional learning infrastructure, and the uptake of quality curricula lag behind the more recent momentum in literacy, despite substantial improvements in curriculum quality and availability. When thoughtfully designed and integrated, AI could serve as a coherence accelerator–strengthening alignment among curriculum, professional learning, and instructional expectations.

That is why our team at Leading Educators is increasing action research on how AI can be integrated responsibly into instructional systems—not as a standalone solution, but as a support for the professional learning routines educators already use.

One promising example is a new multi-year, multi-district learning pilot that we’re calling the MathAI Impact Network, developed in partnership with Playlab with financial support from Gates Foundation and Valhalla Foundation.

Over the next year, educators across approximately 26 schools in three participating districts–Boston Public Schools, KIPP North Carolina Public Schools, and Richland School District 2–will test AI-enabled supports embedded directly into curriculum-specific math professional learning routines for grades 3-10. We aim to surface how AI can strengthen the efficacy and scalability of professional learning without fragmenting instructional priorities or increasing the burden on educators.

Albert Kim, Chief Innovation Officer at Leading Educators, shares:

Our goal isn’t to automate teaching or push for AI use because it’s trendy, as is the case with many edtech-driven initiatives. This technology is already entering classrooms. We want to help systems critically understand, optimize, and integrate safe AI tools within a vision for math excellence to amplify the human work of teaching and learning.“

Most efforts to bring AI into schools begin with tools. This work starts elsewhere: with instructional priorities, leadership practices, and the professional learning structures that districts already use. Only then do systems ask where AI can help strengthen coherence, save time, and support better instructional decisions.

Britney Wray, a Senior Director of AI-Enabled Instruction who is leading the project, adds:

The real work of improving instruction happens through people—through coaching conversations, collaborative planning, and leaders making thoughtful decisions together. If AI has a role to play, it has to strengthen that human work, not distract from it.”

About the Work Underway

The math pilot sits within a larger body of AI-focused innovation work that Leading Educators is co-designing with school systems, educators, curriculum publishers, and support organizations.

Our collective charge is to ensure AI strengthens what teachers and students do best, anchored to three core principles:

• Think deeply. Use AI to unlock, not replace, critical and creative thought.
• Connect meaningfully. Use data and design to foster belonging, not isolation.
• Learn faster together. Model what it means to adapt with purpose and integrity.

When taken together, our AI pilots are designed to provide clear signals on how best to anchor AI in the instructional routines and learning systems we already know drive results.

The MathAI Impact Network will contribute knowledge by exploring how AI-enabled supports can strengthen lesson internalization, instructional planning, and coaching in alignment with HQIM. These are recurring responsibilities for teachers and instructional leaders, but ones that are often constrained by limited time, uneven support, and inconsistent structures. The first year will introduce five tools developed by Leading Educators’ curriculum, content, and coaching experts towards this end:

1. Curriculum internalization tool: Enables educators to deepen their understanding of the learning objective and progression of learning through activities and discourse in a lesson.
2. Discourse tool: Enables practice and feedback on facilitating mathematical discourse.
3. Assessment tool: Enables teachers to identify students’ specific learning needs and then create small groups to ensure students reach the lesson objective.
4. Instructional walkthrough tool: Enables leaders to give content-specific, curriculum-aligned feedback to teachers.
5. Coaching tool: Enables leaders to prepare for coaching conversations.

From the start, educators and leaders have been positioned as co-designers and sources of insight—helping shape both the tools developed by Leading Educators and the professional learning conditions required for responsible use.

The work launched in November with classroom observations, empathy interviews, leadership sessions, and foundational professional learning. Over time, the partnership plans to expand to additional systems and apply increasingly rigorous research methods to study impact, implementation, and cost.

What This Will Mean for the Field

These early insights point to a clear signal: AI alone does not fix fragmented instructional systems. But when embedded in coherent professional learning, it can help educators and leaders do the work they already know matters.

For districts considering similar efforts, the most important questions are not about features or platforms, but about:

• Instructional priorities and coherence
• Leadership practices and expectations
• Time, routines, and support for educators
• Guardrails that ensure AI strengthens—not distracts from—teaching and learning

Over the next two years, this collaboration will continue to generate evidence through more rigorous studies, including randomized comparisons and cost analyses. The partnership is committed to sharing open-source resources, research briefs, and practical guidance so that other systems can learn and adapt.

For now, sharing early learning responsibly is essential. The field does not need another promise about what AI might do. It needs clear signals about what it takes for AI to support strong instruction in real schools.

As learning deepens, our team at Leading Educators plans to refine how this work is named, shared, and expanded—anchored in what educators and systems find most useful.