HAT

Research has demonstrated the effectiveness of human high-dosage tutoring to improve student learning outcomes and close achievement gaps.
However, a central challenge remains: How might we scale human tutoring while maintaining effectiveness and quality of service?
Our key insight is to blend what AI does best - finding patterns in large volumes of data - with what humans do best - nurturing and caring for others.
As part of the HAT project, we are designing a hybrid human-AI solution to scale the effectiveness of human tutoring.
Our team uses ASR and discourse classification models to analyze tutorial sessions for use of research-based practices.
These practices are backed by the academically productive talk (APT) and belonging-centered instruction frameworks.