Learning with peers helps students reflect, generalize knowledge and apply it more successfully to new problems. How can we scale successful peer learning from the controlled environment of the small classroom to the wild, massive scale of online classes? In my talk, I will introduce computational systems that structure peer learning at massive scale, and demonstrate their efficacy through the results of randomized controlled experiments with more than 10,000 students.
I will show how insights from educational theory can be distilled into interfaces that scale the classroom practice of peer review to online classes and enable accurate feedback on open-ended work like essays, programming, and visual art. I will also show how, combined with the scale of an online class, carefully designed interfaces can yield improvement-oriented feedback on open-ended work in just twenty minutes, enabling students to revise and learn to mastery. Similarly, our Talkabout system assigns students to small geographically-diverse video discussion groups in real-time, demonstrating how global diversity can promote reflection and a deeper understanding of concepts.
In classes across disciplines including computer science, psychology, and design, more than 100,000 students on Coursera and EdX have used these systems for peer assessment and discussion. These wide deployments point to a future where systems recognize the large scale of users and their diversity as a computational resource that enables novel applications in learning and creative work.
Chinmay Kulkarni is a PhD candidate in Stanford University’s Computer Science department. In his research, he has created tools and techniques that have enabled students to learn better through interactions with peers, both in online classes and at several universities. Chinmay served on the Program Committee of the new ACM conference on Learning@Scale, and was recognized as a Siebel Scholar in 2014.