Khanmigo is an AI-powered, chat-based tutor that personalizes activities for students. By combining the existing chat functionality with data about learners and classrooms, Khanmigo balances providing grade-level practice with closing prerequisite gaps, while supporting teacher and student choice. This will give students the support they need to accelerate their rate of mastery of grade-level skills.
Khan_Academy-Logo

Project Title

Khanmigo: Equitable AI-powered learning at scale

 Mountain View, CA
Where it is used:
  • New Jersey
  • Indiana
Project Summary
 

What’s the problem that Khan Academy is trying to solve?

Khan Academy seeks to connect the lessons teachers are teaching in their classrooms to resources that help students mastering both the skill itself and its prerequisites. Too often, teachers must identify the unfinished learning of their students and assemble materials that support students in learning prerequisite skills. The workload required to do this for all students across all of a teacher’s classes is immense, and leaving this task to be done by individual teachers is neither scalable nor sustainable.

What does Khanmigo do?

In March 2023, Khan Academy launched Khanmigo -a chat-based tutor that is aware of a student’s knowledge and skills to help guide students through math practice. In addition to the AI-powered tutor, Khanmigo includes a set of activities built with learning science principles. The solution will allow teachers to leverage the AI recommendations to effortlessly diagnose needs and assign personalized content work related to their current topic. As students engage in this work, data from their progress will be used to uncover prerequisite skills that the learners have not yet mastered and which are likely to impede their progress on grade-level content. These skills will be recommended to learners and teachers as ways to unblock their progress and support grade-level work.

What is the wow factor?

Khanmigo’s solution will use data from student performance to recommend the best next steps for learning on our platform. Simultaneously supporting student agency and choice, Khanmigo will balance student, teacher, and system-driven recommendations by empowering teachers and students with the ultimate control over accepting or overriding recommendations. The team will build features to support students to develop the metacognitive skills needed to make these decisions.

How does it work?

Students interact with a chatbot that is powered by a large language model to help them work through math practice. Using information about the student’s proficiency and preferences, Khanmigo passes that information into the conversation to personalize the experience and provide intelligent recommendations, prompting students to engage.

or contact Barb Kunz at barb@khanacademy.org

Team Members

Kristen DiCerbo
Chief Learning Officer at Khan Academy
Dr. Kelli Millwood Hill
Created the Efficacy & Research Function at Khan Academy