q&ai

Overview

q&ai is driven off a simple and familiar assumption: Teaching things makes you understand them better. Our application, modeled off a tutor-student system, asks the student to tutor the computer, which follows up with questions about predicted gaps of understanding from the original answer. We use machine learning models to analyze what students say and cross-reference them with concepts extracted from source data, inputted when content creators were building the original curriculum.

Today, we're pursuing a gap in industry with education targeting the medical professional education industry. We leverage our AI models, educational research and strategic partnerships to drive content created by hospitals and consumed by medical students, residents and newly hired clinicians. In this vertical, we improve quality of medical practice, education and analytics. Currently, we are actively running pilot programs, and if interested, please reach out!

Demo

q&ai - Medical Demo.mp4




To the left, you can find a demo video exploring the functionality of our Alpha release of this application! Reach out if interested in trying this application yourself!


General Information




The research behind this project began at the University of Michigan. This research was formalized in the paper A Scalable, Flexible Augmentation of the Student Education Process, which was presented at the NeurIPS 2018 AI for Social Good Workshop.

108-A_Scalable__Flexible_Augmentation_of_the_Student_Education_Process.pdf

Research

As an organization, we are focused on enabling education with Artificial Intelligence. This requires development of new algorithms and models. Some of the active areas of research that q&ai continues to pursue are:

  • Follow Up Question Generation - At the moment, a teacher (when building curricula) provides a question title, and source text. Based on just this, we should be able to use recent research on question generation to provide questions to the teacher (as supplemental or discussion questions) that also have to do with the source text.
  • Student Answer Alignment - Grading students is currently a very simple process - we'd like to bring semantic analysis into grading, and try to provide students autogenerated, ground-truth answers to help them understand where their misunderstanding may lie.
  • "Justify Your Answer" - Recent discussions with some educators have highlighted such a tool's use in STEM fields, which our prototype is not currently built for. In STEM fields, it's often not enough to give an answer; rather, the learning happens when you have to explain why.

Contact Information:

Bhairav Mehta [CEO] - bhairavmehta95[at]gmail.com

Adithya Ramanathan [CTO] - adithya.ramanathan[at]gmail.com

Michael Staunton [Head of Engineering] - mstaunton19[at]gmail.com