Joachim Lebovits, Sorbonne Paris Nord University/PyxiScience and 
Jacques Lévy Véhel, PyxiScience

Title: PyxiScience: AI-assisted adaptive mathematics learning and grading platform

Abstract:

The PyxiScience platform offers an innovative solution for teaching mathematics, particularly suited to contexts characterized by heterogeneous classes and teacher overload. Co-founded by Joachim Lebovits (Sorbonne Paris Nord University/New York University) and Jacques Lévy Véhel (former research director at INRIA), PyxiScience launched a year and a half ago in France and the United States and now has more than 10,000 users.
One of its main features is AI-assisted correction of handwritten papers, which reduces grading time by a factor of ten while making assessment criteria more objective. The tool provides detailed analyses of students' skills beyond simple grades, identifies gaps in reasoning, generates personalized exercises, and creates dynamic, needs-based recommendations. Under supervision, AI contributes to proposing content and supporting the teacher in remediation until mastery is achieved.

We will discuss how this platform can be integrated into a mastery-based grading sequence. This presentation aims to spark a discussion on the contributions of AI to mathematics education.