Project Activities
The team will develop the ARIS prototype across three iterative cycles. In the first cycle, developers will add a written explanation field to G3MS challenge questions and build an initial AI model that classifies student responses as reflecting a misconception, a foundational gap, or low confidence. In the second cycle, the team will add AI tutor conversations as a data source and build a teacher dashboard that displays classifications and recommendations. In the third cycle, the team will use ARIS classifications to route students to their next lesson. Throughout development, researchers will compare ARIS classifications to those of trained human raters to assess reliability, and students and teachers will provide feedback on usability. At the end of the project, the team will conduct a pilot study with approximately 150 to 200 students in grades 5 to 10 in partner schools in Brooklyn, New York and Broward County, Florida. Students will be randomly assigned to G3MS with ARIS, G3MS without ARIS, or a retrieval practice comparison condition for six weeks. The team will examine whether the prototype functions as intended, is feasible for teachers to use, and shows promise for improving student mathematics learning.
People and institutions involved
IES program contact(s)
Project contributors
SubAwardee(s)
Products and publications
Project website:
Questions about this project?
To answer additional questions about this project or provide feedback, please contact the program officer.