Perusall: a symbiosis between machine learning and pedagogy

Abstract

In this paper we present an exploratory study that aims to contribute to a better understanding of artificial intelligence, particularly machine learning potentiality in pedagogical practices, both in terms of student engagement and teacher’s work namely in monitoring and students’ assessment. We used Perusall, a collaborative reading tool that analyzes student interactions with content and with other students, through a quality algorithm, providing various reports and assigning individual grades according to 6 parameters. Perusall, integrated on moodle of the Universidade Aberta, was used in the context of a 2nd year curricular unit, of the Education undergraduate program. An activity inspired by inquiry-based learning, lasting 1 month, was designed, where it was intended to read a 43-page document, create 3 questions, answer 2 questions from colleagues and vote on the 3 best questions (based on quality parameters provided by the teacher). After this activity, students answered a questionnaire about Perusall and pedagogical strategy. 246 students were involved (referring to 2 school years), of which 204 completed the activity. The results, based on Perusall data and student reports, show high levels of motivation that could mean deeper reading and more consolidated learning, further contributing to greater self- and co-regulation. From the professor’s perspective, artificial intelligence facilitates individual and collective monitoring and makes the assessment process faster and more objective.