ProductiveMath
ProductiveMath leverages carefully sequenced challenges—where learners attempt complex problems before instruction—to deepen conceptual understanding and transfer. Our work explores how AI can generate, curate, and scaffold Productive Failure tasks while keeping assessment stealthy and formative.
Project Description
What is Productive Failure?
In a PF sequence, students first tackle non-trivial problems with minimal guidance, surfacing prior knowledge and misconceptions. Targeted instruction follows, yielding durable learning.
Our Platform: ProductiveMath
An AI-powered authoring and delivery tool that (1) generates diverse PF tasks, (2) detects strategies via process data, and (3) serves just-in-time scaffolds without spoiling struggle.
Research Questions
- How well do AI-generated PF tasks elicit target strategies and misconceptions?
- Can stealth signals (attempts, sequences, time) model productive struggle?
- What supports keep struggle desirable for varied learners?
Project Details
- Lead: Dr. Seyedahmad Rahimi
- Sponsor: Bill & Melinda Gates Foundation
- Funding: $451,563
- Duration: 2024–2026
Publications
-
Rahimi, S., Ercan, D., Gao, R., Esmaeiligoujar, S., Babaee, M., et al. (2025). ProductiveMath: A Generative-AI-Powered App to Support Productive Failure Teaching. AIED 2025, 344–351. Springer.