NSF grant ~$1.51M: merging math-learning data on 90,000 elementary students into an "AI-ready" research base (Florida State)
A ~$1.51 million ($1,509,240) NSF grant to Florida State University. Six existing research datasets covering 90,000 elementary students and 3,000 teachers are integrated into a shared, AI-ready dataset focused on early mathematics — numerosity and operations such as addition and subtraction. Phase I of the IDEAL-Math plan, run by learning scientists, data scientists, AI researchers, and education practitioners together.
Grant overview (primary data)
- Award amount$1,509,240
- RecipientFlorida State University (Florida)
- ProgramSci of Lrng & Augmented Intel
- Period2026-07-01 〜 2029-06-30
- FunderU.S. National Science Foundation (NSF) / NSF
Key points
- Recipient: Florida State University; ~$1.51M ($1,509,240); July 2026–June 2029 (Phase I)
- Integrates six existing research datasets — 90,000 elementary students and 3,000 teachers — focused on numerosity and early operations
- Surveys, assessments, classroom videos, and think-aloud videos harmonized across overlapping measures into an AI-ready shared base
- Grounded in a model linking achievement to teacher development, knowledge, practice, classroom environment, and student attitudes
- Dashboard and user workshops planned — groundwork for a dataset that actually gets used
Early struggles with arithmetic cast a long shadow over later math achievement and STEM careers, which is why testing "what works" against large datasets matters so much. But education-research data accumulates study by study, in incompatible formats and measures — individually too small, collectively unusable.
This project attacks that unglamorous but decisive bottleneck: entity-matching and integrating scattered research data, at a scale of six studies, 90,000 students, and 3,000 teachers.
1The key term: AI-ready
The keyword is "AI-ready." Progress in learning analytics and educational AI has been rate-limited less by models than by the existence of large data prepared for learning. Only when surveys, assessments, classroom video, and think-aloud video are harmonized across overlapping measures can machine learning surface meaningful patterns — which instructional practices help which students, and when.
The award embodies, in education, the field-wide shift of AI's frontline toward data engineering.
2Bringing practitioners and policymakers into the design
Equally notable is the team design: education practitioners and policymakers sit alongside learning scientists, data scientists, and AI researchers from the start. Defining research questions first and letting them guide the harmonization — plus the dashboard and user workshops — is a hedge against the classic failure mode of databases that get built but never used.
As "Phase I" signals, this is the foundation stage, and its success will determine the larger build-out to follow.
Why it matters
Educational AI and learning analytics are rate-limited by prepared data, not models — and public funding is now flowing to that foundation. For EdTech builders and education-policy readers, how an AI-ready base is designed for K-12 math is a transferable reference for data integration in other domains.
FAQ
What does "AI-ready" data mean?
Why focus on early mathematics?
Is personal data being published?
Sources (primary)
Source: NSF Award Search (U.S. National Science Foundation, public domain). Amounts are the obligated amount. For privacy, we do not handle principal investigator names.
- NSF Award (original, official)
- NSF Award ID: 2621428