AI Assistance for Grading Handwritten Free Responses
Gradescope by Turnitin is a web application that enables instructors to grade their existing paper-based exams and homework, via scanning student work in first. Our goal is to save the instructors as much time as possible with AI-assistance. For example, we group answers to fill-in-the-blank questions by content, such that instructors grade each unique answer only once, instead of grading the same answer many times, as on paper. In this talk, Sergey will share recent work on AI-assistance for short-answer questions, where several sources of complexity prohibit an easy solution. First, student handwriting can be messy and often mixes math and science notation and diagrams with text. Second, grouping similar answers together is a known challenge in natural language processing. Third, grouping answers also depends on instructor intent and their grading rubric, which tightly links potential ML solutions to UX interfaces.
Sergey Karayev’s goal is to develop and deploy AI systems to improve human life. In 2014, he finished a PhD in Computer Science at UC Berkeley and co-founded Gradescope, an AI-assisted platform for instructors to grade exams, homework, programming projects, and labs. In 2018, Gradescope was acquired by Turnitin, where Sergey now heads development of AI for STEM. Sergey is also an instructor of a weekend program and University of Washington course on Full Stack Deep Learning.