M-Powering Teachers: A Machine Learning Tool for Instructional Measurement and Feedback
Joined 2022
Partners
Stanford University
Harvard University
What are we building?
An automated feedback tool using speech recognition to analyze classroom interactions
A high-quality benchmark dataset of math instructional sessions
A system to deliver formative feedback to teachers based on verbal practices
Infrastructure for research using recordings and automated tools
What are we learning?
How to fine-tune automatic speech recognition (ASR) models for instructional settings
How to analyze classroom discourse to develop measures of instructional talk and feedback equity
How to deliver effective automated feedback through experiments, interviews, and coaching collaborations
Products
Automated instructional feedback tool
Benchmark dataset of instructional audio/video
Research findings on feedback practices and instructional equity
AIMS Collaboratory | Inventory of Public Goods
Public goods shared by the M-Powering Teachers project team:
For: Researchers
The Promises and Pitfalls of Using Language Models to Measure Instruction Quality in Education
Assesses multiple high-inference instructional practices in in-person K-12 classrooms and simulated performance tasks for pre-service teachers using Natural Language Processing techniques.
Academic Article
For: Researchers
Automated feedback improves teachers’ questioning quality in brick-and-mortar classrooms: Opportunities for further enhancement
Presents results from a randomized controlled trial with 224 Utah mathematics and science teachers to assess the impact of automated feedback on teachers’ use of focusing questions.
Academic Article