MTM Dataset: 4,654 math tutoring transcripts open to researchers

Researchers led by René Kizilcec and colleagues from multiple institutions have released the Million Tutoring Moves (MTM) dataset, a first-of-its-kind open multimodal resource for studying tutoring interactions. MTM v1 contains 4,654 math tutoring transcripts collected from a U.S.-based nonprofit online tutoring platform. The dataset is part of the broader National Tutoring Observatory (NTO) infrastructure, which aims to translate authentic tutoring interactions into actionable insights for research, practice, and the development of AI-powered educational technology.

By making these tutoring sessions systematically observable and analyzable, MTM v1 supports research on instructional processes and enables the creation of AI systems grounded in real educational interactions. The dataset is safe, open, large-scale, broad-coverage, and multimodal, setting a foundation for future expansions. This initiative can accelerate the science of tutoring and help build more effective AI tutors that learn from how humans actually teach math.